The company is attempting to stop the cash burn that has put it in jeopardy.

CNBC has obtained an investor deck The Messenger was using as recently as late December to entice potential individuals or companies to infuse it with $20 million.

The Messenger, which started in May, launched on the idea of becoming a down-the-middle digital news juggernaut. It initially planned to hire around 550 journalists and generate over $100 million in revenue in 2024, according to The New York Times. The company ended up hiring a staff of 300 people and has since struggled financially, which has led to some recent layoffs, according to multiple reports.

The Messenger ended 2023 with a net loss of $43 million, according to the documents. The deck tells investors that with the infusion, the company plans to end 2024 profitable, with net income of $13 million.

The Messenger confirmed to CNBC that the deck was part of a “draft presentation,” and said there have been “adjustments” to the numbers within the documents and that the company intends to “make $13 million and be profitable in 2024.”

“It should also be pointed out that our traffic is growing at an enormous pace. Comscore latest numbers show that we generated 88 million page views in November, and Google Analytics shows that we generated 100 million page views in December. Our traffic is growing at 30% a month, already putting us ahead of many major news publications,” the company spokesperson said in a statement to CNBC.

The documents say that The Messenger is planning to eliminate 40 positions and furlough 15 people for four months this year amounting to an estimated $6.2 million in annual savings.

That’s one of the details that’s since changed, according to a spokesperson. The company laid off about 25 people last week to save cash, as first reported by The New York Times.

“The layoffs impacted two dozen people, not 40, which was one of the adjustments made to the presentation,” the spokesperson said.

Betting on advertising turnaround The immediate turnaround will be based on what could be an insurmountable climb in advertising sales. In 2023, The Messenger took in $2 million in direct ads and $1.8 million in programmatic advertising. This year, The Messenger forecasts it will bring in more than $18 million and $37 million for each, respectively.

“By 2024 The Messenger will be a known brand in the United States which users will know and make part of their daily media consumption habit,” the company says in its investor deck. “The attention paid to media in 2024 is expected to be very high. We have a critical U.S. Presidential Election in 2024 with political and related news content in high demand as well as news events such as debates, primary voting, and conventions.”

While U.S. companies are counting on political advertising to boost sales in 2024, digital media companies that rely on advertising have been ravaged for years by Google , Facebook and Amazon , which have sucked up available inventory. This has crippled companies such as Vice Media and BuzzFeed, which grew too quickly amid advertising revenue declines.

The Messenger will be relying on Google search to drive programmatic advertising. On the direct side, $10 million of The Messenger’s forecast $18 million will come from Messenger TV, a yet-to-be-launched service that will require 19 additional employees, the presentation shows.

Most of the Messenger’s expense has been head count; it spent about $39 million in 2023 to hire hundreds of employees.

Despite the hope of a financial turnaround, the deck indicates that there is no plan for The Messenger to cut back on millions of dollars in spending. For instance, with the creation of Messenger TV, overall personnel expense will rise to more than $48 million in 2024.

The Messenger expects to have open its three facilities in New York, Washington, D.C., and West Palm Beach, Florida, according to the deck. The facilities payments are estimated to exceed just over $240,000 each month this year.

Travel, meals and entertainment expenses at The Messenger are estimated to be more than $1.7 million by the end of 2024, with the company expected to spend over $140,000 each month of 2024.

The Messenger highlighted the severity of its cash problems and illustrated the tough sell it will have to make to investors for more money.

The company had negative cash flow of $3.8 million in October, according to the deck. It then added $5 million in November and an additional $1.7 million in “incremental investment” to stem the cash burn.

But the business has already incinerated the incremental investment in two months, the deck says. The Messenger ended December with $667,000 in cash. It plans to end January with monthly cash burn of $4.2 million, pushing the company into negative cash territory by the end of the month.

While The Messenger plans for the advertising market to turn later in 2024, it acknowledges the business will likely hemorrhage cash in the coming months.

Without additional investment, The Messenger predicts, its ending cash balance by June will be negative $16 million. The company predicts operations will generate positive free cash flow in August.

From the recruitment interface of Apple's official website, we can see that Apple is also recruiting talents with experience in running ChatGPT and other areas as well as conducting basic research on Large Language Modle (LLM). According to the New York Times, this indicates that Apple is carrying out long-term AI research projects that will affect the future of Apple and its mobile terminal products.

Apple's Machine Intelligence and Neural Design (MIND) teams are actively recruiting researchers and engineers working on AI and LLM in areas such as Seattle, Cupertino, Paris and Beijing.

While OpenAI and Google, among others, are slightly ahead of Apple in the field of generative AI, this hiring by Apple is also a sign that Apple is working hard to bring generative AI technology to its mobile products.

How to combine generative AI with mobile terminal products is also a technical problem that has not yet been solved by OpenAI and Google and others. And the solution to this problem is critical to Apple's hardware sales and software services business.

In addition, according to the foreign media, the New York Times reported that Apple is recruiting dozens of positions working on large language modeling in its offices in California, Seattle, Paris and Beijing. After searching the official website, Wisdom found that there are more than 600 positions related to big language modeling in Apple's recruitment.

01 Apple's AI Layout: Increasing R&D Spending, Improving Terminal Software Efficiency

In a conference call with investors last Thursday, Apple CEO Cook said that AI and machine learning are Apple's core technologies. These technologies are an integral part of Apple's production of almost every product.2023 Apple's Q2 earnings report showed that Apple's R&D spending was $3.1 billion (about RMB 22.3 billion) higher than the same period last year. Cook said that part of the spending was mainly on generative AI, and said that Apple's R&D spending on generative AI will increase further in the coming months.

Previously, Apple's external stance on applying generative AI has been more cautious.

According to Lukasz Olejnik, an independent researcher and cybersecurity consultant, Apple's short-term goal is to apply generative AI technology to existing products to enhance the experience of using existing products.

Olejnik further said that the application of generative AI technology on cell phones can improve the operational efficiency of applications, handle user data in a safer way and protect user privacy.

In addition to the application of generative AI technology, Apple will also be machine learning (Machine Learning) technology to further protect user privacy.

02 Apple Recruits AIGC Talents Widely

In terms of Apple's talent recruitment, some of the job requirements focus on optimizing the multimedia capabilities of AI software, including generating image or video content, visual actions and interactions.

In addition to machine learning, Apple is also looking for people with a "deep understanding of large-scale language modeling and AIGC," and promises to apply the technology to iPhones and other devices. Apple hopes that AI technology can be applied to "how people communicate, create, connect and consume media".

Apple Labs in Paris, for example, has already hired some researchers from Meta, according to a French AI entrepreneur who has worked for a large tech company, and the lab plans to recruit more AI talent in the future.

In terms of company acquisitions, in 2020, Apple had spent nearly $200 million (about 1.4 billion yuan) to acquire Seattle-based AI startup Xnor, whose main business is to run complex machine learning models on mobile devices, and which had beaten out large companies such as Microsoft, Amazon and Intel.

Apple declined to comment on its hiring plans.

03 CONCLUSION: AI may be a new growth area for Apple in the future

Whether based on external pressure from other companies, or for internal consideration of Apple's own product optimization, from the existing news, Apple will make efforts in the field of AI, and will use AI technology mainly for product software service optimization. Cook's statement on the second quarter earnings call also indicates that Apple will continue to increase R&D expenditures in the field of generative AI in the future.

The arms race in the field of AI not only includes tech giants, but also star startups. Who can continue to survive and win in this race? What kind of changes will Apple's move bring to its products and services? Are other tech companies using this to launch new AI end products? All these are still unknown, and we will continue to pay attention.

"Entering the AI era, Huawei Pangu big model will come to help Hongmeng ecology." On August 4, Huawei executive director, terminal BG CEO, intelligent car solutions BU CEO Yu Chengdong introduced, through the underlying technology of Pangu big model, Harmony OS brings the next generation of intelligent terminal operating system.

The use of large models on cell phones has long been nothing new, and previously ChatGPT, Wenxin Yiyin, Myo Duck, and other APPs and small programs have been used to meet the needs of AI applications for mobile terminals by calling the cloud computing power.

The next step is to let the big model run directly on the cell phone.

Since April and May this year, the American technology giants - Qualcomm, Microsoft, Nvidia, the most popular AI star OpenAI, as well as the domestic AI "head team" of Tencent, Baidu, etc., have accelerated the promotion of AI big models in mobile terminals. Lightweight deployment of AI models. Qualcomm even announced that it is gradually transforming into an intelligent edge computing (providing computing services at the source of data such as mobile terminals) company.

Under the strong push of the group of giants, the industrial trend of big models moving from cloud to end has been very clear.

Why should the big model "run" on the cell phone?

The biggest feature of big models is "big", often tens of billions of billions or even trillions of parameters, and in order to better run big models, the arithmetic cluster has been upgraded to the level of "10,000 cards". Now, why is it necessary to "stuff" the big model into a small palm-sized cell phone?

Large models will indeed bring some experience improvements to cell phone users. For example, Huawei terminal intelligent assistant small art can not only recommend restaurants according to voice prompts, but also summary summary, information retrieval, multilingual translation and other information processing, thousands of words of English long text, with the ability of large model of the cell phone intelligent assistant can generate a summary, but also can be translated into Chinese. Especially the latter point, in the era of information explosion, for improving the efficiency of learning is still very valuable.

Jia Yongli, President of Huawei Terminal BG AI and Intelligent Full Scene Business Department, explains that, on the one hand, the Big Language Model has the ability to generalize, which can help the cell phone intelligent assistant to improve its comprehension ability. On the other hand, the plug-in capability of the big model Plug-in can open up the barriers between applications within the cell phone and expand the capabilities with the help of tools.

In addition, AIGC applications such as ChatGPT have always been accompanied by strong privacy and security controversies, but this problem can be completely avoided if they run completely on the end-side. Because the big model runs on the end-side, the data doesn't leave the end-side either. Moreover, the response will be faster this way.

On the other hand, the need for big models to mobile terminals such as cell phones has become very urgent.

The surging momentum of big models makes the cloud increasingly unable to carry the demand for arithmetic power on its own. Alex Katouzian, senior vice president of Qualcomm, recently said bluntly, "With the accelerated growth of connected devices and data traffic, overlaid with climbing data center costs, [we] can't send everything to the cloud."

Not counting data transmission to consume network bandwidth, storage, as well as hardware and other large resources, just the cloud arithmetic has now made the relevant vendors a bit unbearable. chatGPT is only in the reasoning stage, a conservative estimate of the cost of arithmetic in the amount of $ 10 million per month.

The biggest problem is not "expensive", but "lack of".

Previously, even OpenAI founder Sam Altaman has revealed that GPUs are in short supply, and even said that he did not want too many people to use ChatGPT. Recently, there are industry insiders who speculate that the large-scale H100 clusters of both small and large-scale cloud providers will soon run out of capacity, and that the trend of demand for H100s will continue at least until the end of 2024. Current NVIDIA H100 capacity is also heavily constrained by the supply chain.

Therefore, the cloud and terminal to form a cooperation, cell phones and other terminal idle arithmetic resources are utilized to solve the "centralized" arithmetic and "distributed" demand mismatch, has become a large model development "cost reduction and efficiency gains It has become a definite trend in the development of large models to "reduce costs and increase efficiency". What's more, compared with the limited number of center nodes, numerous mobile terminals are called "capillaries" that touch thousands of scenes, which determines that this entrance will be the key to accelerate the application penetration of large models.

How to put the big model "into the pocket"?

"Compared with traditional PCs or servers, the biggest challenge for mobile terminals is how to balance the experience and energy consumption, which is one of the most important core points of the Hongmeng kernel design." Gong Ti, president of Huawei's terminal business software department, emphasized.

The large model requires a large amount of computing and storage resources, especially based on the existing hardware configuration of the cell phone, which requires the software system to be well coordinated to improve efficiency and reduce energy consumption.

Now the phone in order to improve performance, at least 8 chip cores, it is necessary to cell phone system to do the coordination, this process will consume a lot of arithmetic power. If the use of heterogeneous resource scheduling, you can efficiently coordinate CPU, GPU, NPU. Gong body said that the scheduling efficiency can be increased by more than 60%.

Cell phone system can be computing, scheduling the smallest unit is called a thread, the traditional operating system is often tens of thousands of threads running at the same time, there will be a large number of invalid threads. For this point, it can be more lightweight concurrency model to deal with concurrent operations, reduce the invalid thread switching on the consumption of computing power. According to Gong Ti said that the concurrency model can make the task switching overhead savings of 50%.

In addition, in the operating system's task scheduling, which is also the most basic element affecting the smooth experience, compared to fair scheduling, dynamic priority scheduling will greatly reduce energy consumption. Dynamic priority scheduling is similar to an intelligent transportation system, according to road conditions and traffic flow, dynamically adjust the traffic signal light status, such as when the traffic flow in a certain direction increases, the signal light in that direction will turn green in advance, it will reduce congestion and delay.

However, to make the big model deployed to the phone, can still run, just cell phone operating system upgrade and improvement is not enough.

As the big model predictions become more accurate and the network gets deeper, the memory capacity consumed by the neural network has become a central issue. At the same time, it also involves the problem of memory bandwidth, network operation, memory, CPU and battery will be consumed at a rapid pace, which is definitely the weight of the current cell phone is difficult to bear.

Therefore, before deploying to cell phones, it is necessary to compress the large model to reduce the demand for inference arithmetic. However, it must be ensured that the original performance and accuracy remain largely unchanged.

Quantization is a common and important compression operation that reduces the memory space occupied by the model and improves inference performance. It is essentially converting a floating-point model into an integer model, since integer operations are more accurate and faster than floating-point operations.

Currently, quantization technology has also been accelerating breakthroughs. The model trained on the server generally uses 32-bit floating point operation (FP32), and on the cell phone side, Qualcomm has quantized and compressed the FP32 model into an INT4 model, achieving 64 memory and computing energy efficiency. Qualcomm's implementation data shows that after training with Qualcomm's quantization perception, many AIGC models can be quantized to the INT4 model, which improves performance by about 90% and energy efficiency by about 60% compared with INT8.

Large model compression technology is undoubtedly a key factor for AI giants to win the mobile terminal battlefield. This also explains, to a certain extent, why NVIDIA in February this year "quietly" acquired the master compression of large model technology of artificial intelligence startup OmniML.

Large model forced terminal hardware upgrade

"This year we will be able to support generative AI models with 10 billion parameters running on cell phones." Qualcomm senior vice president of product management and head of AI Ziad Asghar, on the other hand, recently said to the public that a model with 10-15 billion parameters can cover the vast majority of AIGC use cases. If the device can already support this parameter level, computing can all be done on the device and the phone will become a true personal assistant.

However, the current generation of flagship version of the cell phone chip can be carried to run 1 billion parameter level large model, Qualcomm in June this year, the top academic conference of computer vision CVPR, successfully demonstrated running on the Android system on the large model, but only 1.5 billion parameters.

Parameters jumped almost ten times, running to the mobile terminal of the big model has stepped on the "gas", that the phone will have to accelerate the upgrade in order to cope with.

Cell phone hardware urgently needs to be revolutionized in the AI gas pedal and memory.

First of all, the larger parameters of the big model, need more memory and storage space to store the model parameters and intermediate results. This requires mobile terminal memory chip capacity, as well as memory interface bandwidth are upgraded.

Second, larger parameters inevitably require more powerful computation and inference capabilities to process input data and output results.

Although, AI gas pedals (e.g., various NPU IPs) on current cell phone chips are almost standard, the design is basically for the previous generation of convolutional neural network design, and is not fully targeted at large models.

In order to adapt to large models, AI gas pedals must be able to have larger memory access bandwidth and reduce memory access latency. This requires some changes in the interface of the AI gas pedal (e.g., allocating more pins to the memory interface), as well as corresponding changes in the on-chip data interconnect to meet the needs of the AI gas pedal's memory access.

One of the important reasons why Qualcomm can call out "10 billion parameters within a year to run the phone" is that it has the second generation Snapdragon 8 processor equipped with the fastest and most advanced AI engine in the history of Qualcomm, compared with the first generation of Snapdragon 8 processor, the AI performance has been increased by 4.35 times, and the energy efficiency has been increased by 60%.

Of course, the training and reasoning of large models with ultra-large scale parameters, even in the cloud, there is an urgent need to break through five walls: memory wall + arithmetic wall + communication wall + tuning wall + deployment wall, the phone has to break through one layer at a time.

However, from "intelligence" to "artificial intelligence", for cell phones, the opportunity is greater than the challenge.

"The impact of the innovation cycle on consumer electronics is more important, and can even lead an industry out of the economic cycle." Zhao Ming, CEO of Glory Terminal, judged that the current smartphone industry is in a new innovation cycle opened by AI and 5G+.

From as far away as a classmate who is breastfeeding a child in the 18th line, he asks, "I heard that ChatGPT can write its own scripts, will you be replaced in the future?" From an investor who has handled tens of millions of transactions in the primary market to an investor who asks, "Can your product be combined with ChatGPT?" Then, at a young age, they saw an advertisement in their circle of friends and asked about the ChatGPT course to improve their future workplace competitiveness, asking, "Can I learn how to keep me from losing my job without raising my salary?"

Some people embrace ChatGPT, some people fear it, some people still do not know why it continues to be hot, some people rely on the call ChatGPT big model to get $125 million in financing, and some people take advantage of ChatGPT to make the company's ARR (Annual Recurring Revenue) to $10 million.

So is the myth of ChatGPT's "wealth creation" true? With ChatGPT not open to mainland China and Hong Kong, how can domestic entrepreneurs ride on this new wave of entrepreneurial opportunities? Will China have a latecomer advantage in the battle of the big models?

ChatGPT ignites the AI boom

ChatGPT is not the end game, it's the beginning of the detonation of AI.

Once ChatGPT was launched, users around the world rushed to it, and within two months, the number of users exceeded 100 million, setting a record for the fastest application to break 100 million on the Internet, a figure that took TikTok nine months, Facebook two years, and Instagram 30 months.

A devastating third technological revolution kicked off. The philosopher of science Kuhn once proposed that the development of science is not a smooth linear development model, but will be like walking steps, every other stage on a step, just as ChatGPT was introduced, pulling artificial intelligence on a new ladder.

OpenAI CEO Sam Altman predicted that AI big model technology, will become the biggest technology platform in the future after the mobile Internet, and the development of chatbots as the interface, together with multimodal models such as images, music, and text, will give birth to world-class large-scale enterprises.

Microsoft CEO Satya Nadella, the enabler behind this, also came out and said that the development of AI at this stage can already be described as "exponential".

The retired Masayoshi Son also publicly said that will end the relatively dormant state all in artificial intelligence. "I am most interested in the artificial intelligence revolution, I believe that humans will be surpassed by computers or artificial intelligence, we hope to become the leader of the artificial intelligence revolution!" Also said that he is a heavy user of ChatGPT and also talks with OpenAI CEO Sam Altman frequently.

ChatGPT is not the full picture of AI, but just a form of generative AI. The prediction from Sequoia Capital is that the wave of innovation in generative AI applications inspired by large models will be trillions of dollars in economic value.

The primary market is also awash in generative AI startups. On the foreign side, according to PitchBook data, venture capital investment in AIGC has grown by more than 400% in the more than two years since GPT-3 was released, reaching a staggering $2.1 billion this year.

Domestically, between November 2022 and April 2023, AIGC financing reached RMB 2.24 billion, equivalent to 41.4% of the total financing in the past two years.

After YUE, a startup gas pedal under Sequoia China, opened registration on February 28, there were 300-400 AIGC themes among 2,000 registered projects; on June 3, the Spring 2023 Startup Camp Roadshow of Qiji Creation Forum, 41 of the 60 participating companies were AI themes, doubling compared to last year, and 39 of them were related to large models. This shows that the attention of the primary capital market to the AIGC track has increased steeply.

Overall, the number of domestic generative AI startup projects and scenarios are far less abundant than those abroad.

An overseas AI entrepreneur told Titanium Media Venture Capitalist, "First, it is more convenient for overseas entrepreneurs to call ChatGPT interface, which is easy to fine-tune and develop new products; second, the open atmosphere of technology innovation in Silicon Valley allows the sky to be the limit, and product applications in overseas markets are more diverse; third, the atmosphere of open source exchange community similar to Reddit is strong. People will put their understanding of ChatGPT, and new applications up for public discussion and re-creation."

On the contrary, in China, ChatGPT is not open for registration in mainland China and Hong Kong, so how can entrepreneurs and major manufacturers get on the "third technology revolution" train?

"Money for big models, no money for application layer"

Entrepreneurship is not only about dreams, but also about the amount of money in your pocket. Building a big model arithmetic, data, algorithms, talent, capital reserves a can not be missing, but this is not the average small and medium-sized startups can afford the cost.

For example, arithmetic resources, data show that the total arithmetic power consumption of ChatGPT is about 3640PF-days (with one thousand trillion calculations per second, it needs to calculate 3640 days). It is said that 10,000 Nvidia A100 chips are the threshold of arithmetic power to make good AI big models. At present, there are no more than 5 companies with more than 10,000 GPUs in our manufacturers, among which, there is at most 1 company with 10,000 Nvidia A100 GPUs.

For example, according to the report "How much arithmetic power is needed for ChatGPT" by Guosheng Securities, it is estimated that the cost of one training of GPT-3 is about $1.4 million, and for some larger LLMs (Large Language Models), the training cost is between $2 million and $12 million. With ChatGPT averaging 13 million unique visitors in January, the initial investment cost is about $800 million, with daily electricity costs in the $50,000 range.

Costly as it may be, the Chinese version of ChatGPT is on a mission. "China will definitely have its own big language model, which is determined by the current situation." Xu Siqing said to Titanium Media Ventures.

China's big manufacturers have already fist-pumped into the state of the 100-mode war. Baidu, Ali, Huawei, Tencent, 360, Shang Tang, Jingdong, KDDI, Byte Jumping and other giant manufacturers have been involved in the research and development of large models in conjunction with their own business and strategic layout, and strive to be the first "China ChatGPT", which is recorded in the history of China's artificial intelligence.

Leonis Capital said that China seems to be the only country that develops its own model infrastructure except for the US and UK monopoly. First, four of the top 10 developers in the world in terms of number of models are Chinese companies, including tech giants such as Baidu and Alibaba, top universities such as Tsinghua University, and government-sponsored labs such as the Beijing Institute of Artificial Intelligence (BAAI); second, in addition to model training infrastructure, Chinese developers are becoming independent at the hardware level, and in response to the threat of U.S. sanctions Chinese companies are increasingly turning to homegrown GPUs, such as Huawei's Ascend 910; third, while Chinese models are about a year behind the top Western models, it is only a matter of time before they catch up.

Xu Siqing agrees: "Open AI is riding high, and their technological breakthroughs have given them an absolute lead in the AIGC field. Chinese big models are rapidly following, following the path explored by Open AI, rapidly iterating and catching up; American companies are standing more in the more cutting-edge exploration field, which is determined by the characteristics of the two countries' companies and the resources at hand."

"We also have a view that the one who gets the talent gets the model, and the one who gets the model gets the world. The models here include pendant models trained for specific scenarios, both ToC and ToB models trained and deployed for specialized scenarios. We are more optimistic about this field than big models, for the simple reason that 98% of the computing time spent on training a model is spent on pre-training, but the next three steps of supervised finetuning, reward modelling and reinforcement learning only use a small percentage of computing time. " Xu Siqing said.

"One does not have ample funds, two does not have the cutting-edge technology talent pool of small and medium-sized enterprises, early to explore opportunities in the application layer, to create some first-mover advantage, is a good choice." A Silicon Valley investor said to Titanium Media Venture Capitalist.

"I think China's big model spell arithmetic is not suitable for general entrepreneurs, understand artificial intelligence entrepreneurs have a great opportunity, because do not have to start from the most basic model brutal arithmetic, than the accumulation of knowledge, team engineering capabilities and iteration speed, on the shallow surface layer of product innovation, such entrepreneurs is a descending blow. Of course, technology is important, industry experience and the choice of application scenarios are equally important. Our seed round investment in HiDream.ai (wisdom of the future) project founder, the former vice president of Jingdong, the Canadian Academy of Engineering foreign academician Mei Tao is based on this consideration." Xu Siqing said.

In addition, Xu Siqing also pointed out that the breakthrough of AIGC technology brings three levels of opportunities to Chinese entrepreneurs. First, it is the opportunity of LLM, big factory or national team; second, it is the opportunity to do superficial application or product-driven application with the help of big model, no matter To B or TO C, there are very many opportunities, characterized by the difficulty of judging which ones are easy to win, as difficult as foreseeing which one must be successful at the beginning of APP emergence; third, it is to establish the pendant model to solve the application level problem, which is in ToB and ToC There is a very wide range of prospects, because the threshold is high and the certainty is relatively stronger. They are characterized by strong modeling ability, and can work on the model to solve the problems that cannot be solved by product innovation.

Finally, he emphasized that there will also be many opportunities for infrastructure-based services that accompany various language models and applications, such as private deployment, cross-cloud, and cross-model service tools.

Foreign big models, AI applications, as much as the carnival

In Silicon Valley, Open AI wants to stand alone, Google, Meta, Amazon can not promise.

Recently, Amazon, which has been invisible, released the Amazon version of ChatGPT, a new model named Amazon Titan. The Titan series of models are divided into two kinds, one is a text model for content generation, and the other is an embedding model that can create vector embedding for creating efficient search functions, etc.

Meta CEO Mark Zuckerberg, not according to the rules, directly hit the "low price", with open source declared war on Open AI "hegemony". On June 16, according to The Information, Zuckerberg is considering commercializing a new version of the open source large language model LLaMA, allowing other companies to use LLaMA for free to develop related AI software and allowing developers to use these products for commercial purposes.

OpenAI has announced plans to create an epic LLM "app store" that will open up all ChatGPT applications, allowing the platform to be bi-directional, allowing developers to shelve their products built on OpenAI technology, and companies to use commercially available LLMs on demand, completely capturing the AI model The platform can realize two-way connection, developers can put up their products based on OpenAI technology, and enterprises can use commercially available LLM on demand, so as to completely seize the AI models.

"The big model competition in the United States must be blossoming, and the degree of openness will become higher and higher, the history of IT development constantly proves this. From Unix to Linux, from iPhone to Android are examples. The leading open power is not enough, and the vendors who are slightly in a passive position will use openness to construct their own ecology, which is a very positive phenomenon for the industry." Xu Siqing told Titanium Media Venture Capitalist.

Looking at the application layer again, there are not a few cases that have gotten commercial validation. Stable Diffusion, a free open source startup, got $100 million in financing in just one month, and its valuation soared to $1 billion; Jasper.ai, a SaaS company that builds applications based on GPT-3 API, was founded 18 months ago and received $125 million in Series A financing, valued at $1.5 billion; Copy.ai, which helps companies write promotional copy. ARR of $10 million in only 2 years.

"There are differences in overseas projects compared to domestic projects. First, the ability of overseas underlying large models is more mature (OpenAI API); second, the commercialization environment is better, users are willing to pay for high-quality single-point tools, which is also the result of experiencing the development of SaaS industry education; third, the ecology of overseas projects is better, there is no squeeze similar to the domestic Internet majors, and there is exploration of new products and new business models in both C- and B-sides. " Shen Huaxin, head of AI research at Yun Cui Capital, said to Titanium Media.

According to PitchBook statistics, the investment circle invested a total of $1.37 billion (equivalent to about RMB 9.369 billion) into generative AI companies in 2022, almost as much as the sum of the past five years.

Not long ago, Mercedes-Benz integrated ChatGPT into the car, opening a three-month test, which showed that its voice assistant could not only complete simple commands, but also carry out continuous multi-round conversations, with a large improvement in comprehension and response quality.

RISO, Skyworth, Azera and other automakers then swarmed in, using cutting-edge GPT capabilities to take the car's intelligence to the next level. The car machine has completely changed from the initial "radio" to a feature-rich intelligent terminal, and with the GPT "brain", it has started to change from a mute, chicken-hearted machine to a driving partner.

And human-vehicle interaction is not the end of AI on board, autonomous driving is the future. The AI model evolutionary upgrade, so that car companies see the opportunity.

It is becoming a mainstream trend to let AI take the initiative to perceive and make decisions, and give up the reliance on high precision maps. A few days ago, Ideal Auto opened the internal test of City NOA (Navigation Assisted Driving), which uses the BEV (Bird's eye view) big model as the main solution, allowing cars to imitate human "brain circuit" driving. Through continuous learning, Urban NOA can also be trained to "chauffeur" users on their commuting routes.

Taking over the Internet, AI is making deeper transformation of the car, and the big guy with four wheels is becoming more and more like a Transformer.

The car machine + GPT Mercedes-Benz first

A "metamorphosis" from the inside out is sweeping the automotive circle, from traditional fuel power to new energy, from driving tools to intelligent products. Over the years, driven by technology, cars have been changing their appearance and interior, and after the Internet has transformed cars, artificial intelligence is here again.

Mercedes-Benz is leading the way in this new wave by transplanting ChatGPT into the car.

On June 16, Mercedes-Benz's 3-month ChatGPT test program was launched in the U.S. It has partnered with Microsoft to integrate ChatGPT into the car engine through Azure OpenAI service. Owners can choose to use ChatGPT through the Mercedes me app, and there is a more intuitive way to test it - use the voice command "Hey Mercedes, I want to join the test program" directly in the car, and Mercedes' MBUX infotainment system will automatically connect the voice assistant "Hey Mercedes" to ChatGPT.

In the past, Hey Mercedes could provide information such as sports and weather, answer questions about the vehicle's surroundings, and control the user's smart home, all of which were standardized. This is ChatGPT's watchword.

Currently, only about 900,000 MBUX-equipped Mercedes-Benz vehicles in the U.S. can test ChatGPT on a priority basis, and Mercedes-Benz intends to use this initial testing period to gain insight into requests from users to determine future development priorities and adjust launch strategies for different markets and languages.

On the matter of accessing ChatGPT, Mercedes-Benz gave an emotional statement, "Everything is aimed around redefining your relationship with Mercedes." Mercedes-Benz wants ChatGPT to reshape the human-vehicle interaction experience, and the analogy is more like the car machine "coming to life" from a mute, function-oriented machine to the role of an in-car life partner.

After Mercedes-Benz, domestic automakers are the first to follow suit.

On June 19, Ideal Auto launched its self-developed cognitive big model "Mind GPT", which was developed by Ideal's spatial algorithm team. It is said that the start date of the big model training was long before the release of ChatGPT. Based on 10 TB of original training data, Mind GPT uses 1.3 trillion Tokens for base model training and can recognize voice patterns and voice content, as well as understand dialects, while providing travel planning for car owners, and even having functions such as AI painting and AI computing.

RISO revealed that after the release of Mind GPT, the ideal car will have a new LUI (user language interface) interaction, "For example, if you want to eat a hot pot, you just need to call the ideal student, and the car interface will generate a picture of the hot pot for you to choose, and then the travel route will be calculated automatically."

Skyworth Auto also recently announced that two of its models, Skyworth EV6II and Skyworth HT-iII, have integrated ChatGPT in their smart car machines. In addition, four car companies, Great Wall Motor, Azera Auto, Xiaopeng Auto and Chery Auto, all applied for GPT-related trademarks last month.

GPT on board has become a trend. According to Junyi Zhang, Managing Partner of Auld Lang Syndrome, the access of GPT technology can enhance the human-machine interaction ability of the car and the interaction ability of comprehensive environmental issues. In the future, car companies will have less and less brand differences in terms of hardware in the same price segment, and when it is difficult to produce much difference in comfort, safety, power and range competition, volume intelligence becomes an inevitable choice.

Intelligent cockpit with a "brain"

ChatGPT is another bookable entry in the evolutionary history of automobiles. The most cutting-edge natural language processing model is applied to human travel tools, and a richer in-car life experience will emerge.

Looking back to more than 30 years ago, the car entertainment function and car computer intelligence, is still a blank. The first generation of car machine was born in the 1980s and 1990s, when people's general concern is still the car's engine, chassis and transmission "three major", suddenly there are some models not only can listen to the radio, but also can swallow the tape, free to play music, the car has some of the shadow of the second living space.

The second generation of the car is added to the DVD playback, MP3, entertainment highlights at the same time, the car and a step towards the driving experience, joined the car navigation. At this time, the solution to the problem of "road blindness" has become a mainstream trend. Many old drivers will remember that in the era of no car network, the Kaleidoscope car navigation has become the standard for high-end models, which uses GPS satellite positioning and map package data in the car to achieve relatively accurate navigation accuracy.

However, except for navigation and listening to songs and radio, people at that time did not expect much from the car machine, and the car machine was often not the main factor in deciding whether to buy a car or not.

In the 21st century, electronic digital technology continues to develop, the form of cell phones first changed. Following this evolutionary line of thought, a large screen appears on the car machine and intelligence becomes a new selling point. Car machine based on linux, WinCE, Android and other systems have been adopted by car manufacturers, and then the car can not only free real-time navigation, but also has a panoramic visual system, car driving assistance systems, such as 360-degree image.

When the car is connected to the network, everything by becomes more different. Online movie, road book, voice control, appointment maintenance, remote diagnosis and other functions were added to the car machine, the screen in the center console is getting bigger and bigger, more and more functions, some manufacturers directly in the cab installed a larger display than the tablet, and even manufacturers have recently rolled up the "full screen", even the passenger and rear are to install the screen.

Finally, the concept of the "third screen" is becoming more and more conspicuous, and OEMs hope that the car can become the third generation of intelligent terminals that affect human life, following computers and cell phones. With technology-rich car machine to occupy the user's mind, expand more business models, become the direction of the car companies are now attacking.

Now, the original concept of "car machine" is gradually replaced by "intelligent cockpit". Not only is the car more and more intelligent, but car companies are starting to roll up the interior materials, audio system, lighting system, and Azera has also released an AR glasses to support movie viewing with the car; the ideal L9 is even equipped with a rear refrigerator, making the car a moveable house.

But whether it's the car or the intelligent cockpit, voice conversation has been a relatively lagging feature in development, and considering driving safety, voice control is very necessary.

Over the past decade, almost all car companies and a large number of AI startups have invested a lot in the field of natural language processing, hoping to optimize the voice interaction experience in the car. Many of them can answer simple pre-set commands, such as turning up the temperature and forecasting the weather, etc. Upgrades and innovations revolve around broadening natural language commands, such as turning on the air conditioner to cool the car or turn down the temperature when the user says "it's a little hot.

But to make the car understand more "human" words, such as planning routes in various dialects and even finding restaurants, may not be as effective as the car owner himself using mobile maps and public reviews, richer voice-based human-vehicle interaction fell into a bottleneck, until ChatGPT appeared.

Natural language large model products (ChatGPT, Wenxin Yiyin, Tongyi Qianqian, etc.) are directly open for C-sides to use, so that developers of intelligent cockpits see the light. The powerful comprehension and logical reasoning ability is expected to make the car machine a driving assistant with hidden business possibilities.

For example, the car owner can tell the voice assistant, "help me find out the destination near the group purchase discount, rating more than 4.5 hot pot restaurant, there will be five people dining, give me a reservation location, and then look at where to park convenient." Put in the past, the car machine absolutely can not understand so much information at once, but for ChatGPT, this is just its base manipulation, as long as there are enough real-time data sources, the possibility of needs being met can be infinite.

The addition of GPT does not just make the conversation smoother, but gives the car a "brain" that can not only answer questions, but also understand needs and generate answers. As for how high the IQ, how fast the response, depending on the car manufacturer's ability to train a large model on board, as well as the courage to "kryptonite" on more bull hardware (chip).

AI how to make the self-driving "brain circuit" more like people?

The richness of life in the car makes the car gradually grow into a carrier full of warmth, it is no longer a boring, cold means of transportation, but a comfortable living space.

The evolution of AI-led cars is not only about GPT, but also about the promotion of self-driving technology, which is even more significant.

The traditional approach to autonomous driving research is to cover all possible driving scenarios by collecting large-scale driving data and testing longer driving miles to ensure that the car has a predetermined response plan in the event of an unexpected situation. However, the complexity of unexpected situations is often unpredictable, and once the system does not have a plan to deal with a particular unexpected situation, driving safety will be greatly threatened.

This is why current assisted driving systems must require the driver to hold the steering wheel in order to deal with real-time emergencies. The learning capability of AI will potentially change this.

Not long ago, a research team from Tsinghua University proposed the self-driving "Trusted Continuous Evolution" technology, which is based on dynamically evaluating the AI's trustworthiness for learning training to ensure that self-driving cars can continuously improve their driving ability from basic active avoidance when encountering new and unfamiliar scenarios, and achieve better driving performance while ensuring safety. Better driving performance.

Simply understood, using AI, self-driving capable cars can actively learn and become familiar with various newly encountered scenarios for continuous evolution, and with the accumulation of driving mileage and data volume, performance is continuously improved.

Ideal Auto is using AI big models for autonomous driving. on June 17, Ideal announced the opening of urban NOA (navigation assisted driving) internal test and will open commuter NOA function to users in the second half of the year. Unlike conventional solutions, RISO adopts BEV (Bird's eye view) big model to sense and understand the road structure information in the environment in real time, so that the car can better imitate the operation habits of human drivers.

In the past, most of the assisted driving systems on most cars used a high precision map solution, which is equivalent to feeding the road conditions to the autonomous driving system in real time for it to make decisions. However, in complex urban roads, there are always areas that cannot be covered by high precision maps and cannot be updated in time, which becomes a major drawback of this solution. With the BEV large model, it is equivalent to AI actively sensing real-time road conditions and making autonomous decisions on driving operations.

Of course, BEV also has disadvantages, for example, at some intersections with large spans and more passing vehicles, the sensor field of view is easily blocked, resulting in the loss of local information as a result of real-time perception at the vehicle end. To make up for this shortcoming, Ideal is purportedly paired with a NeuralPriorNet (NPN) and an end-to-end signal intent network, the former acting as the equivalent of an image reference whenever a car goes through an intersection where a fleet of autonomous drivers has gone; the latter learning how a large number of human drivers react to signal changes at intersections to help the autonomous driving system understand traffic signals.

In response, Zhao Changpeng responded that if the court issues a restraining order, it will only affect CoinAmerica (Binance.US). User assets are safe and the platform's access operations remain normal; the SEC's filing of an asset freeze injunction is baseless, and the application to freeze Cryptocurrency's assets is driven by litigation advantages rather than a genuine concern for the safety of customer assets. Look forward to defending yourself in court.

On April 25, at the 2023 Aoshi Gravity X Smart New Generation Summit, entrepreneurs and investors brought cutting-edge sharing around AIGC. As mentioned at the beginning, optimists believe that after the demographic dividend disappears, the digital people created by AIGC will replace the carbon-based labor force and truly serve the people. However, the investors were quite calm and said bluntly that they absolutely do not invest in AIGC.

AIGC's Strengths and Opportunities

AIGC was mentioned by Yijia Zhang, the founder and CEO of AIGC, in his speech "1 Anchor Point and 40 Discriminative Formulas", one of which is digital people. With the mature application of AIGC technology, the existing content generation model will be overturned, and users will change from content producers to content processors and creators. In the future society, everyone will have a digital doppelganger, opening the era of silicon-based biology.

This era has already arrived, and at the opening of this conference, Miss B, the digital doppelganger of Zhang Yijia, appeared on the conference screen for the opening introduction of the conference, liberating the physical human being. In December last year, Liu Run, a famous domestic business consultant, revealed that he had already used a digital person instead of himself in a short video, and the digital person had vivid facial features and a rich variety of gestures, so it was almost difficult to tell that it was not a real person. Wu also mentioned in his 2022 year-end show that every company may need a digital person by 2023.

According to Li-Ping Chen, co-founder & senior vice president of Silicon-based Intelligence, the first half of the industrial revolution has completed material equalization, while the artificial intelligence digital person will achieve technological equalization. In the current society, the law of two-eight prevails, with companies spending 80% of their efforts serving the top 20% of the population, which leads to an uneven distribution of resources. With digital bifurcation, bank account managers can provide full lifecycle and full-scene, full-time services to customers, digital teachers can provide high-end guidance to children in remote areas, and digital anchors can help farmers and small and medium-sized businesses bring goods. In the context of the disappearing demographic dividend and the era of common wealth, digital people will redefine every industry.

The second is the meta-universe. aIGC accelerates the content construction of the meta-universe world and promotes the renaissance of literature in the new era. Zhang Yijia said that one of the bottlenecks restricting the development of metaverse in the past is the efficiency of virtual content production, especially the production efficiency of digital people, and the maturity of AIGC technology provides the best production tools, which will help accelerate the construction of metaverse content ecology with digital people as the entry point. fundamental changes in the production relations of society. Further, it will promote the renaissance of literature in the new era.

In addition, Kunlun Wanwei CEO Fang Han believes that the browser will not be eliminated in the AIGC era, and there are still opportunities. "The browser was the entrance of the search engine in the previous generation, and the next generation must be the entrance of the assistant of general artificial intelligence."

Absolutely no investment in AIGC

Dan He, co-founder & chief brand officer of Black Mirror Technology, believes that AIGC gives rise to imaginative realizations and the whole business will be contented. Just as the magic novel Harry Potter presents, reciting a string of incantations can summon magic. Now with ChatGPT and AIGC tools, inputting a string of promotions can make imagination become reality. In the future, every company will have a digital workforce, with security fences and smart brains that can replace humans, safe, professional and smart on the job.

For example, the content production of games is a kind of PGC production, which has a huge workload and high R&D cost, and the cost of motion capture equipment ranges from several tens of thousands, hundreds of thousands to several millions. With the improvement of AI technology, the threshold of animation production becomes very low, and the market capacity will be opened rapidly.

Zhou Qi, managing partner of Jinshajiang United Capital, holds the opposite view. He believes that now is not the era of insufficient content, but the era of excess content.

"The number or saturation of Internet users is now extremely close to saturation, with 1.067 billion Chinese Internet users and 1.065 billion Chinese mobile Internet users, basically all mobile Internet users. China's per capita consumption of mobile Internet is more than 6 hours per day. In other words, except for work and rest, people's time is fully occupied." Jing Tao, founding partner & CEO of Davu Ventures, said he is definitely not investing in AIGC because the regulation of content in China is very strict and much of the content is very sensitive. No matter what content is produced, it has to follow the domestic situation. Secondly, the cost of producing content is getting lower and lower, the speed is getting faster and faster, and the volume is getting bigger and bigger. What kind of content do the people need and what kind of content is really valuable to us? Now the average online time of Jitterbug users has exceeded 2 hours, and each person brushes close to 400 short videos every day, which has encroached on a lot of our time, and many of them are spam.

"This is only the content produced by people, the future if it is AIGC, may be a day to produce hundreds of millions of times the content today, who look? Is there that much time? Time is fair, there are only 24 hours a day, sleep 6-8, want to work, study, live, fall in love?" Jing Tao is not without reason, for example, in March this year by the AI-generated "Trump arrested" photos circulated wildly, many people believe it at first glance. In China, a woman's subway photo was spread by AI with one click to undress and became a hot search. These are not only misleading spam, but also suspected of provocation and insult. Not long ago, the State Internet Information Office released the "Generative Artificial Intelligence Service Management Measures (Draft for Comments)", which particularly emphasizes that the content generated using generative artificial intelligence should be true and accurate, and measures should be taken to prevent the generation of false information.

That's why we are excited to announce that we are launching a media account to keep our members and followers updated on the latest news, insights, and trends in the animation industry. By subscribing to our media account, you can get access to exclusive content that you won't find anywhere else. From interviews with industry experts to behind-the-scenes looks at the making of your favorite animated films and TV shows, we'll bring you everything you need to know to stay ahead of the curve.

But that's not all – as a special bonus for our subscribers, we are also offering the opportunity to mint an NFT. NFTs, or non-fungible tokens, are a type of digital asset that are stored on a blockchain. They are unique, one-of-a-kind items that can be bought, sold, and traded just like physical collectibles. By minting an NFT, you can own a piece of digital art that is truly unique and one-of-a-kind.

https://opensea.io/assets/ethereum/0x67E47FCA0895a9dea77aD950D93836C76cbE158B/0

To mint an NFT, all you need to do is subscribe to our media account and follow the instructions on our website. We'll guide you through the process of creating your NFT, and you'll have the option to customize it with your own images and text. Once you've created your NFT, you can keep it for yourself, trade it with other collectors, or even sell it for a profit if its value increases over time.

We believe that offering our subscribers the chance to mint an NFT is a great way to reward our loyal followers and give them something truly special. Not only will you get access to exclusive content, but you'll also have the chance to own a unique piece of digital art that is sure to be a conversation starter.

In addition to our NFT offer, we'll also be hosting regular events and webinars for our subscribers. These events will give you the chance to connect with other professionals in the animation industry, learn new skills, and get advice from experts in the field. We believe that by building a strong community of like-minded individuals, we can all grow and learn together.

So, what are you waiting for? Subscribe to our media account today and start exploring the world of animation like never before. With our exclusive content, regular events, and the chance to mint an NFT, there has never been a better time to join our community. We look forward to welcoming you on board and sharing our passion for animation with you!

Just over three months after releasing ChatGPT, OpenAI has personally added more firewood to the big model craze.

In the early morning of March 15, Beijing time, OpenAI announced the birth of GPT-4, a multimodal big model, on its official website. In addition to optimizing the performance of input modality and text length that can be supported by the model, OpenAI upgraded ChatGPT on top of GPT-4 and opened up the API in one fell swoop - the speed of iteration is astonishing.

In this frenzy belonging to big models, the numbers have been numbing. First of all, it is the number of parameters of the model - previously, OpenAI rolled up the number of parameters of the big model to the 100 billion level with GPT-3 (with 175 billion parameters), but soon, PalM-E, the multimodal big model launched by Google on March 6, sat on the " The largest visual language model in history" position.

Next is the company's wildly increasing valuation. According to a report by Dealroom, a global early-stage project data provider, the valuation of global generative AI companies reached a total of about $48 billion, a 6-fold increase in 2 years.

The domestic AI track heated up late, but the valuation of companies soared as fast as ever - Wang Huiwen's AI company "Light Years Away" was valued at $200 million in its angel round. A big model company recently founded by a technology guru, the model demo has not yet been shadowed, and the angel round valuation has made it enter the billion dollar club - and in the windy mouth of the meta-universe, one million dollars seems to be the ceiling of the angel round valuation of domestic startups.

There are also some tangled, negative voices surfacing in the wind.

On the evening of March 2, a posting with the theme of "Why do we feel that the AI in Europe and America is stronger than ours" caused a lot of controversy. The posters compared the AI development environment in China and the United States, and regarded the development of AI in Europe and the United States as "elite education", while domestic AI is "utilitarian education", which is commercialized, and finally came to a slightly desperate conclusion: the fate of human beings is predetermined in the womb. The robot is not immune.

The posting titled "Why do we feel that AI in Europe and the US is better than ours". Source: Weibo @ChenYiRan - Duke University, post reproduced by them

The violent aesthetics of big models may not be the best choice for most companies to go all out. Algorithmic power, high-quality data, and a high density of algorithmic talent are all expensive tickets needed to get to the big-model table, and most players in China cannot have the equivalent of OpenAI's reserves overnight.

But the rich data dimension and broad application scenarios are the rich mines left to Chinese entrepreneurs by the last wave of Internet that lasted for more than 10 years. In the past month, many small enterprises with scenarios and user data have trained small models adapted to their own business based on the base of large models at home and abroad. A company with a large model reserve of tens of billions of parameters has also "slimmed down" itself and launched a lightweight model for finance and advertising to carry out a new round of data reserve.

At present, polishing the sharp edge of algorithm with small models and making technical reserve for the research and development of big models may be a path for Chinese entrepreneurs to achieve a bend in the future.

"Big model of "all-rounder" vs. small model of "expert

How to make AI smarter and more human-like is essentially a question of education.

For a long time, people were keen to send AI to "specialist schools" to learn the ability to solve specific problems - and small models, often with less than a million participants, were born. Google's AI company DeepMind, for example, put AlphaGO through a refresher course on millions of moves played by professional human players, culminating in a 4:1 victory over Go master Lee Sedol in 2016.

But the downside of specialist education is also clear, with most of the smaller models suffering from bias. For example, when faced with writing marketing copy, small models who are adept at image generation hit a wall. At the same time, the educational resources of the specialist are scattered, and each mini-model needs to be trained from scratch separately for basic training.

In 2017, Google invented a new way of education: Transformer model.

In the past, in "specialized education", AI learning relied heavily on human labeling and selection of learning materials, for example, AlphaGO's learning materials came from professional players, not from children who attended Go hobby classes. The essence of the Transformer training method is to let the AI "focus" on the learning materials of different subjects by itself through a lot of pre-study.

The more data used for training, the more effective the model is at pre-reading; the more parameters, the more accurate the model is at highlighting. The self-focused education method frees human hands while allowing the AI to multi-task different subjects, enabling cross-disciplinary knowledge accumulation.

In 2018, Google released the first basic model BERT with over 100 million parameters based on Transformer and outperformed models bred in the neural network training (such as CNN and RNN) model by far in the subject of translation.

Since then, Transformer has swept the model education community, and the "big" of the big model has been rolled up by many companies. At present, the number of 10 billion participants is considered by the industry as the inflection point for model capabilities to achieve a leap forward.

The most intuitive superiority of big model is that it has the reasoning and deduction ability which is difficult to be reached by small models, and can understand more complex and broader scenarios.

Besides the content production field, where else can big models be used? Li Tao, founder of APUS, a mobile Internet service provider, also gave an example: 80% of the traffic congestion in first-tier cities is not rooted in too many vehicles, but in the intelligent transportation system with low collaboration - how many seconds are set for the traffic lights at each intersection? How do the traffic lights of different road sections cooperate? These problems are difficult to solve with only people or small models.

The emergence of the big model, so that the huge amount of traffic data has a use, "people can only make decisions based on a section of the traffic situation, while the big model can see more comprehensive.

The greater potential of large models is also the ability to reduce the cost of training small models. A large model is like a child who has gone through compulsory education, so it is less expensive to go to college and choose a major, and then become a higher-order professional.

This also means that having a large model as a base from which to train lightweight models for specific application scenarios can save the process of developing basic understanding from zero. The risk of this approach is, of course, that the capabilities of the big model will directly affect the quality of the cultivated models.

What is the difficulty of developing big models?

Over the past month, a number of ChatGPT-like conversation applications have been introduced to the market.

Just from the daily conversation experience, the difference between each product does not seem to be very big. Problems such as fooling or pleasing the questioner and poor timeliness are still common, but compared to intelligent customer service that is confined to specific scenarios and answer templates, the conversational bots that are emerging today have initially made people interested in "wanting to continue the conversation".

But when you look deeper into the details of the model's parameters and Tokens, everything becomes less optimistic. There are only a few startups that have developed their own models to reach the scale of 10 billion parameters, and many of those with significant parameters are a bit fishy.

In order to test the ability of the large model, a strategic analyst of an Internet company showed 36 krypton he designed creative writing, news retrieval, logical reasoning and other 300-400 groups of Prompt (question and answer prompts), more than a dozen breakthroughs in the scale of 1 billion parameters "ChatGPT" application to test one by one need It took two to three months to test each of the more than a dozen "ChatGPT-like" applications with over 1 billion parameters.

After testing, he found that most of the products' answer patterns were too similar to ChatGPT: "It's hard not to wonder how much water is in the 'self-research' model."

The answer is obvious to most practitioners, but helpless: not only do you have to spend a lot of money and time on a large model, but you also need a social environment that is willing to invest in it at any cost.

Algorithms, algorithms, data, scenarios, these are the four key elements to run through the big model. The first two are also conceivable difficulties floating on the sea level, especially for small companies.

The article "ChatGPT China Metamorphosis" mentions these soul-crushing questions: to run through a model with more than 10 billion parameters requires at least one month of training with 1,000 GPU cards, and to a certain extent, the talent that determines the algorithm capability is mostly gathered in Silicon Valley or strong large factories.

The difficulties hidden beneath the surface are the industry values that have long been confined to commercial returns.

"Since the reform and opening up, China's economy has maintained a period of high-speed growth for more than 30 years and quickly among the world's top, which has a lot to do with the development of the Internet pulling more industries for rapid commercialization of landing." A practitioner of nearly 20 years in the AI team of Internet companies at home and abroad told 36 Krypton. But the experience of development, has also become the shackles of inertia, "in front of the new opportunities brought about by ChatGPT, we inevitably still use the old perspective of commercial returns to assess."

Many investors also feel that it is not easy to get money in a cool way. Affected by the tough situation of Chinese stocks and the difficulty for companies to go public in the U.S., many technology companies have become conservative and cautious in their attitude toward U.S. dollar funds. And now that government-led funds have a greater proportion of RMB LPs, funds are facing greater challenges in raising RMB.

Caught in the middle, dual-currency funds are facing the dilemma of not pleasing both ends. "Except for a few head funds that are not short of money, most investment institutions are on the sidelines." A dual-currency fund investor said.

Even if a big model is trained, no one dares to conclude that the capital return will definitely come after the "5+2" investment cycle.

On March 2, OpenAI dropped a bomb of uncertainty on the industry by making ChatGPT's API public at a "cabbage price" of $0.002/1000 tokens (about 1 million words/18 RMB). After only half a month, GPT-4 has landed on the track with the posture of a terminator. This is more than enough to make many domestic companies feel: "Roll but."

The first to be hit were companies in the model layer, where model performance had not yet been honed to a level where they could compete with ChatGPT, and lost pricing power.

Changes in the content industry were also inevitable, such as search, design, copywriting, etc. An employee in the Internet search business talked about the overwhelming response to the new technology reform process: "For example, advertising directly linked to revenue, after generative AI access, users may have the right to choose not to see ads; even if you put ads on, the cost of search has doubled after accessing the big model."

The idea of commercial realization, seemingly as simple as prefixing "AI+" on existing applications, is not very clear.

"Hazy beauty", many investors so described the AI track in the last two months on the subject. "In the technology industry, many new technologies start out as thematic investments, investing in an imaginative economy." An investor who has experienced many windfalls such as meta-universe and Web3 told 36 Krypton, "We tend to think that all the current 'AI+' have the possibility of being made, but because of this, the vision (visionary) and business model of the company will be more in the process of seeking financing emphasized."

They had just learned that BuzzFeed News would be publishing a story revealing their identities to the wider world, after they had been carefully hidden from the world (and they thought they were hiding it well).

Speaking at a hotel in downtown Manhattan earlier this month, Solano recalled, "We received a 20-minute warning.

As Solano and Aronow do when they make big or even small decisions about their business, they made an immediate phone call, everyone freaked out, and then planned their next move. "Frankly, we had very real security concerns," Aronow says. He was sitting on a bench in the hotel's courtyard restaurant next to Solano. the bad guys might try to hack into their accounts. People could show up at their homes or do worse than that to them. We didn't know what was going to happen," they both say.

They began deleting personal information from the Internet. aronow recalls deactivating his Instagram, fearing it might contain clues about the location of his home, and then they warned their families of what was about to happen so they wouldn't be targeted as well.

While Aronow's immediate family fully understands why the upcoming article has garnered so much attention, Solano had to explain the specifics to his father. He and Aronow are the creators of the Bored Ape Yacht Club, the hottest NFT project on the Internet. They launched the Bored Ape Yacht Club in April 2021 through Yuga Labs, a company currently valued at $4 billion.

Bored Apes are human apes made up of 10,000 unique digital head-and-shoulder drawings, each with a unique combination of features ranging from common ("boring" mouth) to ultra-rare ("solid gold" fur). Last October, a rare ape sold at Sotheby's for a jaw-dropping $3.4 million. That same month, Guy Oseary, a veteran artist manager representing Madonna and U2, became BAYC's business partner.

Today, boredom apes are everywhere in pop culture, from T-shirts sold by Old Navy to VMA-nominated music videos shot by Snoop Dogg and Eminem. Celebrities like Steph Curry, Justin Bieber, Gwyneth Paltrow, Post Malone and Seth Green all own Boring Apes. Other high-profile holders include Jimmy Fallon and Paris Hilton, who had a controversial exchange about their apes on The Tonight Show in January. (Solano and Aronow say they didn't know about Fallon's segment beforehand and thought it was "very surreal" to go on the show to discuss it.)

Four days later, Ali and Atalay followed suit, tweeting their names and photos of themselves, Atalay saying, "We wanted to have more control over the narrative and make it something to celebrate more than Greg and Wylie.

Now, a few months later, Solano and Aronow are trying to take back control of their own narrative. So this meeting, accompanied by their publicists, finally told their story in full for the first time and publicly addressed the major controversy that has plagued them for over a year.

The "F*CKING EVIL"

During our time together, Solano and Aronow exude a brotherly but guarded atmosphere, "We are the most superstitious people in the world," Solano says. He wore a fawn T-shirt for the interview because Aronow thought he needed "yellow energy.

Aronow wears an amber bracelet that he considers "positive" and hardly eats his cheeseburger. At one point, Solano kindly pokes fun at his friend's eating habits. He says, "Wylie only eats, like, cheeseburgers and chicken tenders. We joked that he had a 'baby mouth'.

I later learned that this was the result of a debilitating disease that kept Aronow bedridden for much of the decade, starting in his early 20s. He is now able to manage it, but it is still unstable and even salads are a threat to him, causing his disease (he won't say what it is) to flare up again.

The pressure is such. As the public face of Yuga now, this seems like it would hurt. The company is growing: In March, it bought two other favorites in the NFT line, CryptoPunks and Meebits, from creator Larva Labs, and shortly thereafter, Yuga launched what is essentially its own cryptocurrency, ApeCoin.

In the days leading up to our conversation, Yuga will host a massive demo of Otherside, an immersive game it developed with UK-based studio Improbable. Only, it's not just a game - it's the beginning of the Web3 meta-universe, open to the masses beyond the bored ape community. yuga will be competing directly with big companies like Meta.

More stressful, however, are the lawsuits. in June, Yuga sued concept artist Ryder Ripps (which is the same incident that made artists like Kanye West and brands like Gucci famous for their collaborations) for trademark infringement, among other complaints, for creating an identical NFT series to Bored Ape in May. The project earned an estimated $1.8 million in profits, according to Artnet. ( Mainstream NFT marketplace OpenSea removed the series).

But Ripps' cottage project (known as RR/BAYC) is only part of the problem. (Ripps told Input that he worked with three other people on RR/BAYC, including Jeremy Cahen, the creator of the NFT marketplace, who was also sued). Since late last year, Ripps has been very public in accusing BAYC of being rife with racist and neo-Nazi symbolism, claims that the BAYC founders deny, saying they were all part of a plan to drum up interest in Ripps' cottage ape.

It's extremely obvious to anyone who knows our history how ridiculous this is," Solano said. That said, it's hard to see how obsessive, malicious, and frankly, how fucking evil the whole thing is," the trolls are.

A serious look swept across Aronow's face as he described the onslaught of online hate they have received as a result of these allegations. It's like this every day," he said.

Quirky partnership

We're really an eccentric pair in terms of background," Aronow says. He's referring to his friendship with Solano. That much is obvious. The first thing you notice is the huge height difference between the business partners: At 6 feet 2 inches, Aronow is taller than Solano.

Aronow has thick black hair and is covered in tattoos (he's embarrassed by the lifelike portrait of author Charles Bukowski on his right arm, which he got as a teenager). His voice is low and booming. Solano, a bald, goatee-haired, soft-spoken man, calls Aronow the perfect motivational "workout buddy.

We fight over every idea, whether it's a simple tweet or an entire NFT project," Aronow says. In fact, our friendship started with a fight. We first met about a decade ago at a Miami dive bar while home on college break, where they began debating the merits of David Foster Wallace's Infinite Jest, which Solano hadn't even read but reflexively hated because his creative writing classmates raved about it. They kept in touch at a distance, arguing about books, movies and ideas, and playing World of Warcraft online together."

A few years later, in 2017, the two started talking about cryptocurrencies. Like everyone else, they tried to make some money in the bull market. But they were most interested in the possibilities presented by the ethereum blockchain, on which people have built decentralized applications, including gamified collectibles like CryptoKitties, where you can buy, trade and breed unique cartoon cats to create more Kitties.

Despite his interest in digital collectibles, Solano didn't purchase his first NFT until early 2021. shortly thereafter, in February, Solano texted Aronow about starting an NFT project of their own. aronow says, "We started ideating immediately. One of the ideas was a public digital canvas, which Aronow shared with his old friend Nicole Muniz, now the CEO of Yuga. She astutely predicted that someone would paint a penis on it.

I was like, "Where would you draw a penis? The answer: on the bathroom wall of a dive bar. And what kind of people would go there?" The kind of people he knows on cryptocurrency Twitter who have made a fortune from cryptocurrency but still just want to play MMORPGs online instead of living the expected luxurious life of a multi-millionaire.

Aronow sent Solano "a whole article" to plan the idea, in which the name "Bored Ape Yacht Club" came up, and Aronow recalls, "Being the great editor that he is, Solano said - 'That's it. That's it.'" Aronow says he and Solano started a limited liability company the next day. (The information BuzzFeed reporters used to humanely identify him was largely related to Solano's address at the time.)

"I would love to talk to other people who have suddenly created something really popular. It was incredibly surreal."

They weren't artists; at the time, Solano was working in publishing and Aronow was unemployed, so they hired a team to execute their idea. Muniz, the founder of a branding consultancy, introduced them to a visual artist known as Seneca, who created the original BAYC concept art based on directions such as "raggedy punk rock" and "a dive bar in the Everglades. Four other artists helped design the original 10,000 apes.

In a Rolling Stone profile earlier this year, Seneca called her compensation for the project "unsatisfactory. The two said they compensated her for four to five days of work roughly equivalent to Solano's then-five-figure salary, and paid Seneca and four other artists $1 million each late last year. (Seneca did not respond to Input's request for comment).

Meanwhile, Solano got in touch with his friends Ali and Atalay, who had met while studying computer science at the University of Maryland. (Solano met Atalay first, when they were both at the University of Virginia - Solano for his MFA and Atalay for his undergraduate degree.) Ali grew up on the West Coast with immigrant parents from Guatemala and Pakistan, and they met while taking English classes. immigrants from Turkey; he says he had a "normal, suburban upbringing," mostly in Washington, DC.

Solano, who is not a programmer, asked if they knew Javascript, which is not a relevant coding language for blockchain. The tech duo quickly learned the correct programming language, Solidity, and Atalay says, "That was the easy part. Because the ERC-721 token standard for NFT programming has long been publicly available, it was easy to use as a template.

The complexity was managing the multiple components of the project: the website, the smart contracts, the token gated community space, and tying them all together," says Ali. "It was a big challenge to do that with just two of us, especially since we were both just learning blockchain programming.

But they did it. Pre-sales and minting kicked off on April 23rd. And on the night of April 30, 2021 (Hitler's death date), as Ryder Ripps and other BAYC conspiracy theorists will point out - the foursome released the monkeys, which cost $200 a pop. Then they went to bed.

Around 3 a.m. on May 1, Ali got a call from Atalay. He says Ali thought something was "terribly wrong," but instead, he and Atalay watched the collection sell out in real time in the early morning hours. ( As it turns out, word of BAYC had spread through the rabid NFT community). At that moment, they knew they had created something big.

This all happened almost exactly 15 months ago. Shortly thereafter, all four co-founders began working full-time at Yuga Labs. Yuga added new NFTs to the ape family, giving them dogs in June 2021 (via Bored Ape Club airdrop) and mutant apes in August; the latter series sold for $96 million within an hour of launch. steph Curry was one of the first celebrity holders, and he sold his first NFT in the same month when he bought his ape for about $180,000.

In September 2021, Muniz joined as a partner (a month before Oseary) and became CEO in January of this year. in March 2022, Yuga Labs raised $450 million in a funding round led by a16z. Today, the company has about 70 employees, and Aronow says, "I'd love to talk to someone else who has suddenly created something that's quickly become so wildly popular. It's just incredibly surreal."

The four founders insist they don't live glamorous lifestyles. They all bought houses, in locations across the U.S. that they refuse to disclose (except for Solano, who lives in Miami). But Solano and Aronow say they spend most of their time in their homes, working at least 10 hours in their home offices - Solano's is unadorned, and Aronow's is covered in BAYC items and children's wallpaper, which he says came with the house .

But they do get to meet celebrities, like Snoop Dogg and Colin Kaepernick. aronow says, "sometimes Guy will introduce us to someone on a speakerphone. oseary will hold his phone up to the screen. He'll FaceTime-with people I totally admire," says Aronow, whose favorite celebrity is primatologist Jane Goodall (Yuga Labs contributes one percent of the total ApeCoin supply to her foundation.) Solano says he enjoys meeting digital artist Beeple.

Perhaps the biggest change for Solano is that he is now uncomfortable taking business calls from celebrities. In the past, he says, he was just as introverted: "I wouldn't want to call and order Pizza Hut.

Outlandish characters

Solano and Aronow, who both grew up in Miami, lament that the city has become less cool because of the industry they work in. Their Miami is not the concrete jungle of their cryptocurrency brethren living in downtown skyscrapers; Aronow says his Miami is particularly "lush, beautiful, and full of very strange characters.

As a kid, he met a lot of these characters - powerful "1980s Miami Vice era dudes" who were old friends of his father's. They would take him out to lunch and tell him stories about his father, Don Aronow, who was murdered in 1987 when Wylie was a baby.

Don Aronow was the son of Jewish immigrants born in Brooklyn who made his fortune in construction in New Jersey in the 1950s and later became a leader in the powerboat industry, making a name for himself in the process, Aronow said. There he witnessed "the emergence of the industry," selling boats to and dealing with "movie stars, kings and queens.

Subsequent President George Bush Sr. was a friend who owned Don's powerboat. U.S. border security agents used his boats - just like the drug smugglers they hunted down.John Travolta played Don in a 2018 film, Speed Kills, which Aronow described as "terrible. RogerEbert.com gave it half a star.

In 1987, Don was shot in his car in North Beach. He was 59 years old. About a decade later, two people (a former powerboat industry competitor and the killer he allegedly hired) said "no contest" to his murder. But there are still plenty of conspiracy theories surrounding the businessman's death, involving mob ties and a jealous partner who was an alleged mistress. Don was also, according to online speculation, a CIA agent.

Whatever the truth about Don's life and death, Aronow says the stories he heard from his father's friends often conflicted with those of his mother (former Wilhelmina agency model Lillian Aronow), whose death deeply influenced Wylie's childhood. He grew up in Coconut Grove with his mother, stepfather and an older brother, eight years older, who was the "bar hero" of the Miami punk scene.

Aronow says he doesn't want to offend his parents, but describes his home environment as "shitty. He spent much of his childhood playing video games, such as Final Fantasy. By the time he was about 12, he was often running away from home to attend local punk rock shows, like his older brother before him. There, Aronow found a kind of 'second family' of similarly troubled characters.

By the age of 15, Aronow was an alcoholic and drug addict. He would run away from home for months at a time, sleeping on construction sites and in the mangroves with other young runaways. He says he has been to the same 'court-ordered treatment' facility twice. He added, "The head of one facility said I was the worst case of alcoholism he had ever seen in a teenager."

He was sent to a second 'very confusing' facility. It's like the kind of place Paris Hilton went to, where they kidnap you at night and take you out into the desert," he says. The people who ran the Utah center wouldn't let him read anything except the Bible or a science book, which is basically the Bible of Alcoholics Anonymous. aronow chose the latter, which he says he read about 50 times while in Utah, and it changed his life.

The moment I got back to Miami, I became Captain A.A., trying to help other alcoholics see the same message I felt in the desert," he says. He was 15 years old at the time, and he smoked and drank black coffee with recovering alcoholics decades older than him (these days, Aronow drinks occasionally, but doesn't do drugs)."

I really wanted to give everything because in the back of my mind, I thought, maybe I'm going to be sick again.

Aronow went to college and had hopes of getting into a top MFA program, like the one at Syracuse University taught by his hero, writer Joe George Saunders. Those dreams were dashed when Aronow became seriously ill in his early 20s and had to drop out of school.

Again, Aronow would not speak of his illness - given his superstitious nature, he didn't want to give it "that kind of energy. But it kept him in bed for much of the 'really dark decade. Because of his illness, his family supported him financially. He traveled the country in search of doctors who could help him (mostly to no avail), he studied meditation and hung out in online communities, 'scraping by' through Twitch streams and YouTubers. He trades in cryptocurrencies, but has never had a real 9-5 job. He somehow managed to meet his current girlfriend.

Finally, in his early 30s, he found the right specialist, the right medication and the right diet. He got better. Then, at the same time, the rest of the world got sick too. The new crown epidemic took over. With everyone stuck at home, people looked online for avenues to get involved in their communities, just as Aronow has done for the past decade. NFT as art and collectibles made a big splash on the scene.

Then, in February 2021, Solano texted Aronow, "Hey, do you want to do an NFT?"

Aronow said, "It was like, I want to do everything. I really want to give everything because in the back of my mind, I'm like, maybe I'm going to get sick again."

Ordinary people

By his own admission, Solano's backstory is far less dramatic than Aronow's. His parents, both Cuban immigrants, came to the U.S. very young, when his mother was a baby and his father joined the Young Communist Pioneers when he was old enough to join the Young Communist League. (Solano only made his first trip to Cuba about seven years ago).

Solano's mother has lived in Miami all her life, and his father moved there in his 20s after serving in the U.S. Army Special Forces, also known as the Green Berets. His parents divorced when Solano was 11, and he and his older sister live with his mother, who works for a television network he declined to name.

Solano has wanted to be a writer since he was 11 years old. After graduating from NYU, he moved south to earn an MFA at the University of Virginia, which he says ended up being "the best time of my life. He was able to spend all his time writing, he quickly made friends, and he met his future wife, now a professional landscape architect, who first came in contact with him because she liked his poetry.

"You see the founders as a larger life character online. Then you meet them and think, well, that really could be me."

After graduating from the University of Virginia, he landed a job at a small press working on licensed intellectual property - Harry Potter coloring books and World of Warcraft manuals. It wasn't a dream job, and it wasn't the highest-paying job (cheap Bored Ape NFTs now sell for twice his salary at the time), but it could be satisfying. When he retires later, he says, "I want to write a physical book, which is amazing."

He realizes the irony of the fact that - his work is now almost invisible. But at the same time, we're always trying to make it more tangible," he says.

Indeed, BAYC has real-life benefits (including access to branded merchandise, such as hats and hoodies). What's more, ape-holders own the intellectual property of their apes, which opens up a wide world for branding. When you talk about people spending $X for any of these boring apes, what's in that spending is the promise, the opportunity, the chance to be the next Mickey Mouse!" says actor Seth Green.

( In May, Green lost his Bored Ape Fred in a phishing scam. Green has since managed to regain ownership of his ape, saying he struck a "multifaceted" deal with the NFT collector who bought it from the scammers, and, he told Input, "The Los Angeles County Sheriff's cybercrime unit is working on the case. Once I made the deal and got my apes back, I was relieved to be able to look at the market with God's eyes. Instead of feeling so crazy, like all my plans are being terminated at the touch of a button.")

Beyond the business, there's the fun that comes in the form of IRL events. Last fall, in conjunction with the NFT.NYC conference, BAYC organized its first ApeFest, which culminated in a warehouse party in Brooklyn with performances by the Strokes, Beck and Lil Baby (an early celebrity Ape holder).

Back then, the founders could enjoy the festivities without being known by the broader BAYC community, says Ali: "I remember last year at ApeFest, I could hang out with anyone and no one knew who I was. Once the leaks happened, I knew that would not be the same." The second ApeFest, during the return of NFT.NYC in June, boasts four nights of performances by acts like LCD Soundsystem, HAIM, Snoop Dogg, Eminem and (again) Lil Baby.

Then there are the more modest BAYC gatherings that take place around the world. For example, the party I attended with Solano and Aronow after our interview. It was held on the leafy terrace of a dark basement bar in downtown Manhattan, where about 15 ape holders, one of whom rolled out his own brand of boring ape hot sauce, gathered for drinks. As is the case with the NFT scene in general, the group skewed young and male.

Plans for Ape Day 2023 emerged. Someone said, "I'd love to go to Las Vegas. I don't know how many apes live in Las Vegas, but ......" Others came up with the idea of bringing the festival to the desert. One prominent community member, Josh Ong, replied, "Apechella!" The concept of the 2025 Tokyo Ape Festival was floated. solano suggested going to "a mountain" somewhere.

For the next hour or so, we nerds were guided through a series of tasks by Curtis and Blue. We searched for a 'new pair of glasses' for a guy who broke them at the recent NFT.NYC conference and banded together to incapacitate the dangerous, drunken Koda (Otherside's 'first boss fight', we were told). At the end of the demo, everyone posed for a set of 'selfies'.

On cryptocurrency Twitter, the reaction to the demo was overwhelming. A few days later, Solano, Aronow and I chatted about the demo on Zoom, and Aronow said, "Honestly, it was one of our best launches. They pointed out that while there were quests to complete and drunken bad guys to topple in Otherside, the point of the game wasn't that it was a game. It's a virtual retreat where players can simply hang out with their ape brethren, buy assets in the world with ape coins and own them in the form of NFTs.

It's an "interoperable metaverse," they explain, because people can bring their NFTs in and out of Otherside and use them elsewhere on Web3. Fortnite, for example, makes money by players buying in-game assets, such as skins for their avatars; its publisher, Epic Games, made more than $9 billion from the game in 2018 and 2019.Aronow says, "All of that value is going into the metaverse, and none of it is coming back out."

It's also a collaborative meta-universe. Voyagers (users who buy with Otherdeed tokens) are given a piece of land in the metaverse that they can use as they wish (according to the Otherside document: "with the informed guidance of the community"). Voyagers can also provide feedback through the Discord server, which will likely influence the shape of Otherside," Aronow explains, "They're iteratively feeding into what this will become at every step along the way.

"The ambition and scale of what we're trying to accomplish here is huge."

This, he continues, will help Yuga compete with Internet giant Meta in the race toward the ultimate metaverse. The ambition and scale of what we're trying to accomplish here is enormous," he said.

Solano and Aronow believe they are the right people for the job, Aronow says, "In the wrong hands (potentially bad actors) a metaverse can be a utopian, scary place. They envision their meta-universe as lush and beautiful and full of very strange characters. ( Although it may be, as someone said in the group chat for the Otherside demo, packed with "a lot of men").

Otherside will take years to develop, they say. But when it's ready, one has to wonder, who cares? The cryptocurrency world is growing rapidly, and by then, won't the apes boom be over? Also, the cryptocurrency and NFT markets have been falling lately.

When I brought up the ongoing cryptocurrency winter at the hotel, Solano told me, "It's the game the losers are playing that concerns the bears." Plus, they're not worried. They have plenty of money.

We are exceptionally profitable," Aronow added. We have a very substantial competitive fund here to make sure we can survive, and not just survive, but build through any multi-year bear market. Just keep building, building, and building again."

It's a huge challenge that the founders could never have imagined when they were just a pair of completely anonymous partners discussing dick drawings on hypothetical bathroom walls. aronow says, "we may be the biggest company in the NFT space right now, but we're far from the biggest company in terms of building a metaverse. We're going to kill some giants."

The humble coffee stand is located outside Seattle's NFT Museum and has only one thing on the menu - a $5 cup of freshly ground coffee.

Creating a coffee flash store to test the operation of a brick-and-mortar DAO DAO brands itself as a "café for everyone" with a mission to open a café run by DAO. To create a new business model that rewards growth and profitability by sharing it with the entire company. Customers and employees are awarded "coffee tokens" that are used to grant governance rights and discounts on each purchase. The token itself is still in alpha stage (Alpha: an internal beta version, not generally released to the outside world) and is not yet connected to any blockchain.

The project's website indicates that the easiest way to earn coffee tokens is to purchase coffee at the project's distribution events. Another thing is to submit a coffee proposal to the Discord #Proposals channel to help improve this project. Currently CP1 - Coffee Proposal 1 has been completed, which is primarily a low commitment solution to (1) test token economics, (2) identify committed members, and (3) establish order among a group of strangers. In addition, the CP2 - Seattle NFT Museum proposal is complete. Seattle NFT Museum opens its first brick-and-mortar store operated by DAO as a temporary coffee shop, with a flash store open 4/30 to 5/01.

TheCafeDAO's opening of the pop-up coffee shop is the latest in a series of DAO experiments that seem to have been launched by the high-profile but ultimately failed debut of the constitutional DAO last November, proving to many the potential and difficulty of what DAO can do in real life.

The idea to create TheCafeDAO was originally proposed in a Reddit post last August by Dan Car, one of the DAO's founding members, who posed, "What if we made a decentralized Starbucks?"

The idea quickly caught on, and the other five core members of the organization all saw the post and volunteered to take on leadership roles to bring the concept to life. While the organization's focus in recent months has been on a flash store in Seattle, DAO's founders hope it will eventually become its own larger, self-sustaining entity.

Because DAO is still in its early stages, the founders also recognize that decision-making is not yet as decentralized, although that may change as the community continues to grow. another DAO founder, Dustin Tong, said in an interview: "Our goal is for Discord to grow big enough so that we don't have to decide the future of DAO ourselves." "Just yesterday, someone brought up the idea of running a hot dog stand with DAO. That would be awesome."

Token Economics Testing phase

Coffee tokens have not yet launched on any blockchain. theCafeDAO is currently testing the token economics of TheCafeDAO by simulating the minting and distribution process using ETH signatures, Google forms and an ad hoc process for surveys. While this is an imperfect representation of the blockchain, this process allows TheCafeDAO to test, learn and iterate quickly and economically before finalizing a solution. During the M1 - M2 testing phase, TheCafeDAO will be minting and distributing iCT (faux coffee tokens) and plans to airdrop real CTs to iCT holders at launch. see the current distribution of iCTs on the Distribution page and learn how to get iCTs on the Earn Coffee Tokens page.

Minting Principles

CTs are also used as governance tokens to manage DAOs. Individuals/groups with more CTs have more voting power over the DAO. CTs must be distributed fairly according to the following guiding principles: incentivize those who are adding value; prioritize those who are actively adding value; and deprioritize those who are no longer adding value.

Coffee tokens are minted when Café DAO generates revenue and burned when fees are paid. The following diagram depicts the minting process.

(Step 1) Customers who purchase a 12 oz. coffee for $5.00 will (Step 2) mint 1 CT (coffee token). The Treasury will keep the CT and $5.00 until the operating costs are calculated. Assuming that the operating costs total up to $3.50 per $5.00 coffee, (Step 3) the Treasury will burn 0.7 CTs and spend $3.50 to cover the operating costs.

(Step 4) Coffee tokens are allocated to the following 3 parties: customers, employees, and DAO. the DAO Treasury will retain its own CTs while ensuring that all CTs in circulation are backed by the appropriate amount of dollars. The allocation of CTs to each party will be decided by a vote of the DAO. Non-token holders can either exchange their tokens for fiat currency or hold them for governance rights, and a cup of coffee equals one vote in any proposal made by DAO members.

The role of CT

1 CT = 1 vote

Token holders have proportional voting rights to manage the DAO. Token holders can vote on the treatment of CTs held in the DAO treasury.

Converting CTs to USD

The DAO treasury will always honor the exchange of each CT for a dollar priced coffee and burn the exchanged CTs. even if all token holders want to exchange their tokens for dollars, this is possible because the DAO treasury will always hold an equivalent amount of dollars for each CT in circulation.

Exchanging CTs for coffee

Since CTs are worth the equivalent of 12 ounces of coffee, CTs can also be exchanged for a cup of coffee. However, CTs cannot be purchased directly from Café DAO. All CTs are minted from sales and distributed through the process shown in "Distribution" above.

It is important to note that Café DAO will always provide 12 ounces of coffee, equivalent menu items or coffee prices for 1 CT. However, DAO will not attempt to control the price of a CT on any secondary market, whether it is above or below the price of a cup of coffee. While it may be relatively stable, no attempt will be made to correct any decoupling from the secondary market.

Fundraising Methods

Coffee Tokens cannot raise funds for direct casting by purchasing CTs. Instead, the Café DAO funding mechanism allocates future DAO CT distributions. In order to be eligible for future CT distributions, individuals need to improve the DAO in one of the following ways: creating a coffee proposal, executing a coffee proposal, or funding a coffee proposal through a loan.

The following is how Café DAO raises money to execute projects that cannot be self-funded (Step 1) Token holders submit a CP (Coffee Proposal) to improve the Café DAO. the DAO will review and choose to pass, reject or request more information from the submitter. If the proposal passes, it will move to (Step 2) the "DAO Consideration Queue" where the DAO will need to determine (1) how much the proposal is worth, (2) the terms of the DAO loan, and (3) how to incentivize the team to execute the proposal. Once the DAO has finalized all the details, (Step 3) the DAO opens the proposal for lender funding and identifies an implementation team to handle the project. Once the project is completed, (Step 4) the repayment period begins. In the example below, the amount needed to fund the proposal is $100. Café DAO will still allocate CTs to customers and employees, but will reallocate DAO's share of CTs to repay the executive team and lender's loan. The Café will need to generate a profit of more than $100 to repay the lender's principal + interest and the executive team's interest.

Preparation of the permanent storefront Although the coffee flash store was only open for two days, members of DAO are now looking to raise funds to create a permanent storefront somewhere in the Seattle area.

TheCafeDAO is not the first DAO to use a physical location as an extension of its online community. NFT (non-homogeneous token) based projects such as LinksDAO and PizzaDAO have similar plans to use the DAO structure for fundraising and running a full-fledged reality operation.