The big models and the AIGC behind them are on fire, a race to the mountains of technical bulls.

Mobile Internet era, the division of front and back office duties, programmers, engineers and other technicians "sealed" in the background, they built the product structure, modified the bugs, optimize the experience, and stand in the foreground to talk about the product manager, the credit to the user experience insights and create a "temple" level products. The product managers are the ones who create the "pantheon" of products based on their insights into the user experience.

Nowadays, the technicians fighting in the front line have become the middle and senior managers of enterprises, or CTO, or the person in charge of a business line, from accompanying the enterprise to complete the growth of 0 to 1, to maintain the status quo, they have long left the front line, but the heart is always "technical people".

The detonation of ChatGPT may be a salvation for the "technical people" who harbor dreams.

The dream of being happy

"Quite nostalgic for the era of staying up all night to write code, and I also hope to find the state from 0-1," this is what Jia Yangqing said publicly after leaving Ali.

When it comes to Jia Yangqing, the industry is familiar with him as the "Framework God". He majored in automation at Tsinghua University and worked/interned at National University of Singapore, Microsoft Research Asia, NEC Labs, and Google Research during his studies, and in 2008, he went to the University of California, Berkeley to pursue a Ph.D. in computer science, and during that time, he created the open-source deep learning framework Caffe, which has been adopted by Microsoft, Yahoo, NVIDIA, and Adobe.

In 2013, he joined the google brain team and participated in the development of the TensorFlow platform under the leadership of the "Turing Award triumvirate"; in 2016, he joined Facebook, during which he created ONNX, the first prototype of the open model format; in March 2019, he joined Ali as the Vice President of Technology. In March 2019, he joined Ali as Vice President of Technology, responsible for technology, products and business in the direction of big data and AI.

Leaving Ali on March 20, 2023 in the company of dystopian philosophy, "I announced my graduation with the team. the 20th of March is the last day of winter, and to borrow a phrase from Albert Camus: the period of every winter is the blossoming of spring flowers."

Some people speculate that it's an AI architecture startup, others speculate that it's a big model. Jia Yangqing said in April at "Technology Morning Know" that he would not do big models, but would do AI application layer building in the To B direction to fill the gap between small and medium-sized enterprises (SMEs) and AI, and believed that this was a once-in-a-lifetime opportunity.

On March 7, Li Mu left AWS to co-found Boson.ai AI with Professor Alex Smola, the "father of parameter servers", who had joined Amazon AWS in July 2016 as a vice president-level scientist.

On March 8, the media broke the news that Li Yan, the head of MMU (Multimedia understanding, multimedia content understanding) of Racer, set up an AI company, Yuanshi Technology, whose main business is to carry out the research and development of multimodal large models.

Li Yan graduated from the Institute of Computing Technology of the Chinese Academy of Sciences, is the core figure of the AI technology research and development of the fast hand, in November 2015, formed an internal deep learning department DL (Deep Learning) within the fast hand, the goal is to build an algorithmic model to identify illegal and illegal video content. As Racer generated more demand for video content understanding. in 2016, the team changed to MMU group.

On March 24, Wang Changhu, the person in charge of visual technology at Byte Jump, has left to join the Longhu Group, forming a startup team to focus on the visual multimodal algorithm platform of generative AI.

Wang Changhu joined the byte jump artificial intelligence laboratory as director in 2017, and the main research direction includes computer vision, video understanding, multimedia retrieval and machine learning. With the internal position transfer, he moved to become the head of commercialized visual technology at ByteDance, and is responsible for a number of major product lines such as Jittery, TikTok, and Today's Headlines.

As executives Li Yan, Wang Changhu in a large enterprise also has the body, followers of the business needs, turned into an assistant for business growth.

"R & D origin in the technology era entrepreneurship is a natural advantage," said an executive to the "Titanium Media Venture Capitalists", AI development is very fast, although large enterprises want to keep pace, but due to the interests of all parties involved in the internal more, the ability to collaborate between the teams to solve the problem is limited, but rather is to hinder the AI innovation ability to exert.

"If I can reopen, I will make the product more innovative, and the market fit a little higher." Jia Yangqing said publicly.

On the other side of the ocean, the same bridge is being played.

Raja Koduri, Intel's chief architect, left the company on March 22 to start a generative AI startup focused on gaming, media and entertainment. He said on social media platforms, "Generative AI rekindled my passion to get into software and hopefully leverage non-CUDA hardware to handle this kind of workloads, which is what I started the business for in the first place."

Lacroix and Lample, who had both been involved in engineering the development of LLaMA, left to found Mistral AI, with products for large-scale language modeling and generative AI, and a $113 million seed round a month after its founding.

Standing at the turn of the times, it seems that their departure reveals the dreams of the intentional.

Dreams and Reality

The bridge between dreams and reality is money.

"The scariest thing about technician startups is to fall into a technical obsession and ignore commercialization metrics, and money is the lifeline of a business, whether it's financing, or self-blooding." Industry investors said.

Technologist entrepreneurship and scientist entrepreneurship have a lot in common. For example, for a particular technology has excellent professional ability, "factory" comes with a halo, past industry resources, for the technology outlook and industry change insight is profound, is the VC investors favorite crowd.

Similarly, the challenges faced by scientists in entrepreneurship will also be reflected in the technical experts. For example, the comprehensive ability of business thinking, management thinking, marketing thinking to be supplemented, technology is to focus on a certain point to do deep, do fine, and entrepreneurship is to realize the technology, bring commercial value, the face of competition in the market, and have the profitability of continuous innovation skills.

First-class technology bulls "hold the thigh". Silicon Valley big model 4 big players have hand in hand allies.

OpenAI backed by Microsoft, the past three years, Microsoft to OpenAI cumulative investment of 13 billion U.S. dollars, the valuation reached 29 billion U.S. dollars.

Behind Anthropic is its alliance with Google, which has injected $300 million in exchange for a 10 percent stake in the company, and Anthropic has made Google Cloud its preferred cloud provider. Founded by OpenAI's former vice president of security and policy, Dario Amodei, Anthropic has been valued at more than $4 billion in two years and has built up its strength against the rising tide of OpenAI. OpenAI's growing power.

Next, Inflection AI joined forces with NVIDIA to build the world's largest AI cluster. Finally, Cohere has entered into a cloud AI deal with Oracle, which plans to sell access to Cohere's large-scale language models to its cloud customers. Cohere was founded by Gome, a former AI researcher at Google and co-author of the paper "Attention Is All You Need. Cohere, founded by Gome, a former AI researcher at Google and co-author of the "Attention Is All You Need" paper, is currently valued at $2.2 billion post-closing.

Second-rate technology bulls fight for speed, the speed of market sensitivity, the speed of product development and the speed of realization, also known as the first-mover advantage.

A serial entrepreneur said that when a new technology comes to the market, there will be academic papers first, and then apps or demos made by technology enthusiasts will appear on the website for technicians to communicate, at which time it should be launched immediately. The premise is that a long-term deep plowing a technology field, can judge which is the technology iteration brings new business opportunities, which is temporarily unable to bring commercialization.

"Dismantle the technology superstition and look at technology from a business perspective; technology is a tool to help companies realize cash and acquire users." The serial entrepreneur said.

Third-rate technology bull eventually became CTO. "Each person has his own ability to strengths and weaknesses, technology-born founders, pre-entrepreneurship is very advantageous, able to make the product quickly, to the product expansion period, the need to attack the market, if the ability to match up, we will choose to help him find a CEO partner. Not to deny the ability of technology entrepreneurs, just stand in the perspective of investment, in order to let the invested enterprises as soon as possible to generate cash flow, healthy and benign development strategy." Industry investors told "Titanium Media Venture Capitalist".

The fourth stream of technical bulls "wandering in the community" and other acquisitions. "The masters are in the people, AI development speed is very fast at the moment, we will acquire innovative AI projects in the 'community'." A corporate war chest investor said.

"There are a lot of applications made by excellent engineers within the open-source community and technology exchange community, and these applications can help large enterprises quickly get through the cold start phase when they try new business. Large enterprises internal project, development, costs need to be approved, evaluation, which is the cost of time, technology wind mouth comes, we do not want to miss, so we will be in the community to dig out good application projects, find the willingness to sell the project communication." A corporate war chest investor said.

The classification of the so-called first-class, second-rate, third-rate and fourth-rate technology bulls does not specifically refer to the differences in the ability of technical experts, but rather presents a diverse state of existence between the dreams and commercial realizations of technical experts in entrepreneurship.

When technologists start their own businesses, the distinction between technology and making money seems to be less clear-cut. Technology may no longer be a sufficient competitive condition, when the technical concept has been verified, the remaining competition will be reflected in the money, application scenarios.

The big models and the AIGC behind them are on fire, a race to the mountains of technical bulls.

Mobile Internet era, the division of front and back office duties, programmers, engineers and other technicians "sealed" in the background, they built the product structure, modified the bugs, optimize the experience, and stand in the foreground to talk about the product manager, the credit to the user experience insights and create a "temple" level products. The product managers are the ones who create the "pantheon" of products based on their insights into the user experience.

Nowadays, the technicians fighting in the front line have become the middle and senior managers of enterprises, or CTO, or the person in charge of a business line, from accompanying the enterprise to complete the growth of 0 to 1, to maintain the status quo, they have long left the front line, but the heart is always "technical people".

The detonation of ChatGPT may be a salvation for the "technical people" who harbor dreams.

The dream of being happy

"Quite nostalgic for the era of staying up all night to write code, and I also hope to find the state from 0-1," this is what Jia Yangqing said publicly after leaving Ali.

When it comes to Jia Yangqing, the industry is familiar with him as the "Framework God". He majored in automation at Tsinghua University and worked/interned at National University of Singapore, Microsoft Research Asia, NEC Labs, and Google Research during his studies, and in 2008, he went to the University of California, Berkeley to pursue a Ph.D. in computer science, and during that time, he created the open-source deep learning framework Caffe, which has been adopted by Microsoft, Yahoo, NVIDIA, and Adobe.

In 2013, he joined the google brain team and participated in the development of the TensorFlow platform under the leadership of the "Turing Award triumvirate"; in 2016, he joined Facebook, during which he created ONNX, the first prototype of the open model format; in March 2019, he joined Ali as the Vice President of Technology. In March 2019, he joined Ali as Vice President of Technology, responsible for technology, products and business in the direction of big data and AI.

Leaving Ali on March 20, 2023 in the company of dystopian philosophy, "I announced my graduation with the team. the 20th of March is the last day of winter, and to borrow a phrase from Albert Camus: the period of every winter is the blossoming of spring flowers."

Some people speculate that it's an AI architecture startup, others speculate that it's a big model. Jia Yangqing said in April at "Technology Morning Know" that he would not do big models, but would do AI application layer building in the To B direction to fill the gap between small and medium-sized enterprises (SMEs) and AI, and believed that this was a once-in-a-lifetime opportunity.

On March 7, Li Mu left AWS to co-found Boson.ai AI with Professor Alex Smola, the "father of parameter servers", who had joined Amazon AWS in July 2016 as a vice president-level scientist.

On March 8, the media broke the news that Li Yan, the head of MMU (Multimedia understanding, multimedia content understanding) of Racer, set up an AI company, Yuanshi Technology, whose main business is to carry out the research and development of multimodal large models.

Li Yan graduated from the Institute of Computing Technology of the Chinese Academy of Sciences, is the core figure of the AI technology research and development of the fast hand, in November 2015, formed an internal deep learning department DL (Deep Learning) within the fast hand, the goal is to build an algorithmic model to identify illegal and illegal video content. As Racer generated more demand for video content understanding. in 2016, the team changed to MMU group.

On March 24, Wang Changhu, the person in charge of visual technology at Byte Jump, has left to join the Longhu Group, forming a startup team to focus on the visual multimodal algorithm platform of generative AI.

Wang Changhu joined the byte jump artificial intelligence laboratory as director in 2017, and the main research direction includes computer vision, video understanding, multimedia retrieval and machine learning. With the internal position transfer, he moved to become the head of commercialized visual technology at ByteDance, and is responsible for a number of major product lines such as Jittery, TikTok, and Today's Headlines.

As executives Li Yan, Wang Changhu in a large enterprise also has the body, followers of the business needs, turned into an assistant for business growth.

"R & D origin in the technology era entrepreneurship is a natural advantage," said an executive to the "Titanium Media Venture Capitalists", AI development is very fast, although large enterprises want to keep pace, but due to the interests of all parties involved in the internal more, the ability to collaborate between the teams to solve the problem is limited, but rather is to hinder the AI innovation ability to exert.

"If I can reopen, I will make the product more innovative, and the market fit a little higher." Jia Yangqing said publicly.

On the other side of the ocean, the same bridge is being played.

Raja Koduri, Intel's chief architect, left the company on March 22 to start a generative AI startup focused on gaming, media and entertainment. He said on social media platforms, "Generative AI rekindled my passion to get into software and hopefully leverage non-CUDA hardware to handle this kind of workloads, which is what I started the business for in the first place."

Lacroix and Lample, who had both been involved in engineering the development of LLaMA, left to found Mistral AI, with products for large-scale language modeling and generative AI, and a $113 million seed round a month after its founding.

Standing at the turn of the times, it seems that their departure reveals the dreams of the intentional.

Dreams and Reality

The bridge between dreams and reality is money.

"The scariest thing about technician startups is to fall into a technical obsession and ignore commercialization metrics, and money is the lifeline of a business, whether it's financing, or self-blooding." Industry investors said.

Technologist entrepreneurship and scientist entrepreneurship have a lot in common. For example, for a particular technology has excellent professional ability, "factory" comes with a halo, past industry resources, for the technology outlook and industry change insight is profound, is the VC investors favorite crowd.

Similarly, the challenges faced by scientists in entrepreneurship will also be reflected in the technical experts. For example, the comprehensive ability of business thinking, management thinking, marketing thinking to be supplemented, technology is to focus on a certain point to do deep, do fine, and entrepreneurship is to realize the technology, bring commercial value, the face of competition in the market, and have the profitability of continuous innovation skills.

First-class technology bulls "hold the thigh". Silicon Valley big model 4 big players have hand in hand allies.

OpenAI backed by Microsoft, the past three years, Microsoft to OpenAI cumulative investment of 13 billion U.S. dollars, the valuation reached 29 billion U.S. dollars.

Behind Anthropic is its alliance with Google, which has injected $300 million in exchange for a 10 percent stake in the company, and Anthropic has made Google Cloud its preferred cloud provider. Founded by OpenAI's former vice president of security and policy, Dario Amodei, Anthropic has been valued at more than $4 billion in two years and has built up its strength against the rising tide of OpenAI. OpenAI's growing power.

Next, Inflection AI joined forces with NVIDIA to build the world's largest AI cluster. Finally, Cohere has entered into a cloud AI deal with Oracle, which plans to sell access to Cohere's large-scale language models to its cloud customers. Cohere was founded by Gome, a former AI researcher at Google and co-author of the paper "Attention Is All You Need. Cohere, founded by Gome, a former AI researcher at Google and co-author of the "Attention Is All You Need" paper, is currently valued at $2.2 billion post-closing.

Second-rate technology bulls fight for speed, the speed of market sensitivity, the speed of product development and the speed of realization, also known as the first-mover advantage.

A serial entrepreneur said that when a new technology comes to the market, there will be academic papers first, and then apps or demos made by technology enthusiasts will appear on the website for technicians to communicate, at which time it should be launched immediately. The premise is that a long-term deep plowing a technology field, can judge which is the technology iteration brings new business opportunities, which is temporarily unable to bring commercialization.

"Dismantle the technology superstition and look at technology from a business perspective; technology is a tool to help companies realize cash and acquire users." The serial entrepreneur said.

Third-rate technology bull eventually became CTO. "Each person has his own ability to strengths and weaknesses, technology-born founders, pre-entrepreneurship is very advantageous, able to make the product quickly, to the product expansion period, the need to attack the market, if the ability to match up, we will choose to help him find a CEO partner. Not to deny the ability of technology entrepreneurs, just stand in the perspective of investment, in order to let the invested enterprises as soon as possible to generate cash flow, healthy and benign development strategy." Industry investors told "Titanium Media Venture Capitalist".

The fourth stream of technical bulls "wandering in the community" and other acquisitions. "The masters are in the people, AI development speed is very fast at the moment, we will acquire innovative AI projects in the 'community'." A corporate war chest investor said.

"There are a lot of applications made by excellent engineers within the open-source community and technology exchange community, and these applications can help large enterprises quickly get through the cold start phase when they try new business. Large enterprises internal project, development, costs need to be approved, evaluation, which is the cost of time, technology wind mouth comes, we do not want to miss, so we will be in the community to dig out good application projects, find the willingness to sell the project communication." A corporate war chest investor said.

The classification of the so-called first-class, second-rate, third-rate and fourth-rate technology bulls does not specifically refer to the differences in the ability of technical experts, but rather presents a diverse state of existence between the dreams and commercial realizations of technical experts in entrepreneurship.

When technologists start their own businesses, the distinction between technology and making money seems to be less clear-cut. Technology may no longer be a sufficient competitive condition, when the technical concept has been verified, the remaining competition will be reflected in the money, application scenarios.