Concerns have surfaced about the seemingly close relationship between Microsoft and OpenAI.

Recently, according to The Information, an internal Microsoft document instructed Azure salespeople to tell customers that Microsoft could offer more services than OpenAI, while OpenAI was playing defense by delaying granting access to Microsoft's product APIs and holding off on providing the latest models.

The news unveils a lesser-known side of the Microsoft-OpenAI partnership. In the eyes of the outside world, the two are a complementary match made in heaven, with Microsoft providing OpenAI with funding, computing power and other resources to enable the latter to continue to invest in technology research and development, while OpenAI's leading large models and AI technologies are fully empowering Microsoft's software product system, allowing the latter to bend the curve and become one of the leaders of this wave of AI.

Of course, the cooperation between Microsoft and OpenAI is still in the honeymoon period, and there are too many places where the two complement each other for a sales pitch to prove anything. But tracing the roots of this minor disagreement reveals that the conflict of interest in this cooperation is not so simple.

What can't be ignored is that there are competing points of business between the two. In the short term, Microsoft and OpenAI can provide technical services to B-side customers who need to build AI capabilities, so there is a direct business conflict between the two; in the long term, with the further enhancement of OpenAI's technical capabilities, ChatGPT is likely to reshape the C-side application ecology in the future, and it is almost impossible for Microsoft not to be affected at all.

To solve these problems, it is necessary for each of them to make concessions and find a middle ground that is acceptable to both sides. After all, Microsoft is not a charity, and will not be willing to become a pedal for the rise of OpenAI; and OpenAI, although it has taken more than $10 billion from Microsoft, it is still an independent entity, and will not be willing to do only behind-the-scenes credit.

01 Potential Competition Undercurrents

Microsoft and OpenAI's relationship is becoming delicate.

According to The Information, the partnership with OpenAI will allow Microsoft's cloud service Azure OpenAI to directly call on OpenAI models, including ChatGPT, Codex and DALL. This makes Microsoft the only company currently using these AI models for chatbots, search engines, and other products. This makes Microsoft the only cloud service provider currently offering big language models, but it also makes some of the features of Microsoft Cloud Services overlap with those of OpenAI's products.

This overlap means that Microsoft and OpenAI sometimes have to "compete" for customers.

An internal Microsoft document instructs Azure salespeople to tell potential customers that "OpenAI offers limited enterprise-class functionality and security and privacy services," while "Auzre OpenAI's services not only offer the same functionality as OpenAI, but also address compliance, data privacy, and security needs. This rhetoric did help Microsoft cloud services attract customers, such as marketing software startup Touchcast, which chose to access the related Big Model service through Azure because of the security and performance guarantees.

OpenAI, on the other hand, is also looking to connect directly with big customers, including Microsoft's rival Salesforce, while at the same time creating a "time lag" by delaying granting Microsoft access to Big Model in order to work with customers. For example, in March, after OpenAI had signed up Snap and Instacart, Microsoft Cloud Services announced a preview of the ChatGPT feature a week later, and after the GPT-4 language model was released, Microsoft Cloud Services was granted access to GPT-4 only after companies like Duolingo and Stripe had already paid OpenAI directly. 4 access.

Currently, many B-side customers have to choose between "contracting directly with OpenAI" and "using the big language model through Azure," with the latter generally preferred by customers who are already Azure users.

It's almost blatantly clear that not only is there a conflict of interest between Microsoft and OpenAI, but that both sides have acted accordingly.

Some people may wonder why Microsoft is so eager to make money by selling APIs for AI big models when it has just received $10 billion from Microsoft and can now do research without worrying.

From the source, OpenAI's strategic goal is to achieve AGI (General Artificial Intelligence) as soon as possible, and the way to achieve this goal is to use a lot of computation to continuously train, optimize, and iterate on the existing big models. This means that OpenAI faces high computational and R&D costs in implementing AGI.

The MIT Research Review reported that "OpenAI is under increasing pressure to commercialize, and founder Altman sees the need to make money to continue his research. So even though OpenAI started out as a purely non-profit organization with a vision of never allowing AI technology to be monopolized by commercial companies, in 2019 OpenAI made a structural change - adding a for-profit entity with a "profit cap" (a return to investors of 100 times or less). OpenAI LP is a for-profit entity with a "profit cap" (100x return to investors).

Some industry insiders believe that this is just a small conflict between the sales departments of the two companies, so there is no need to make a fuss; however, others believe that the two are currently in the honeymoon period of cooperation, so these small conflicts will not affect too much, but the overlap between the two businesses is always a hidden problem. At the same time, from a longer-term perspective, the two companies may inevitably fight for users in the C-tier products in the future.

ChatGPT's user growth is obvious to all. According to Similar Web's data at the beginning of this year, there are about 13 million unique visitors using ChatGPT every day, with a growth rate of more than 100% year-on-year. In March this year, OpenAI even announced the release of a plug-in feature that allows people to call third-party software directly in the dialog box when using ChatGPT, indicating that as the number of visitors to ChatGPT increases, OpenAI is trying to build its own new application ecology through ChatGPT: using ChatGPT as a platform and vertical AI applications as an extension to help users intelligently use ChatGPT. With ChatGPT as the platform and vertical AI applications as the extension, OpenAI is trying to help users to do various things intelligently, which will have a disruptive impact on the future C-application ecology.

After the announcement of the plug-in, the number of unique visits to OpenAI's website soared to 1 billion in March. According to research by marketing agency VezaDigital, OpenAI climbed nine spots in just one month to become the 18th most visited website in the world in terms of online visits alone.

In contrast, although Microsoft said at this year's developer conference that its main products will integrate generative AI features and enhance productivity through Copilot, the problem is that once OpenAI's ChatGPT becomes a super portal for C-tier applications in the future, Microsoft's products will be relegated to a part of the entire ecosystem rather than the main character.

In other words, although Microsoft and OpenAI are now partners and can cooperate with each other in terms of technology, in the future AGI era, if OpenAI has the ambition and goal to build a new underlying ecology through ChatGPT, then the position of Microsoft Windows system will be threatened. At that time, there is still a strong uncertainty as to who will be the most basic and important portal.

Concerns have surfaced about the seemingly close relationship between Microsoft and OpenAI.

Recently, according to The Information, an internal Microsoft document instructed Azure salespeople to tell customers that Microsoft could offer more services than OpenAI, while OpenAI was playing defense by delaying granting access to Microsoft's product APIs and holding off on providing the latest models.

The news unveils a lesser-known side of the Microsoft-OpenAI partnership. In the eyes of the outside world, the two are a complementary match made in heaven, with Microsoft providing OpenAI with funding, computing power and other resources to enable the latter to continue to invest in technology research and development, while OpenAI's leading large models and AI technologies are fully empowering Microsoft's software product system, allowing the latter to bend the curve and become one of the leaders of this wave of AI.

Of course, the cooperation between Microsoft and OpenAI is still in the honeymoon period, and there are too many places where the two complement each other for a sales pitch to prove anything. But tracing the roots of this minor disagreement reveals that the conflict of interest in this cooperation is not so simple.

What can't be ignored is that there are competing points of business between the two. In the short term, Microsoft and OpenAI can provide technical services to B-side customers who need to build AI capabilities, so there is a direct business conflict between the two; in the long term, with the further enhancement of OpenAI's technical capabilities, ChatGPT is likely to reshape the C-side application ecology in the future, and it is almost impossible for Microsoft not to be affected at all.

To solve these problems, it is necessary for each of them to make concessions and find a middle ground that is acceptable to both sides. After all, Microsoft is not a charity, and will not be willing to become a pedal for the rise of OpenAI; and OpenAI, although it has taken more than $10 billion from Microsoft, it is still an independent entity, and will not be willing to do only behind-the-scenes credit.

01 Potential Competition Undercurrents

Microsoft and OpenAI's relationship is becoming delicate.

According to The Information, the partnership with OpenAI will allow Microsoft's cloud service Azure OpenAI to directly call on OpenAI models, including ChatGPT, Codex and DALL. This makes Microsoft the only company currently using these AI models for chatbots, search engines, and other products. This makes Microsoft the only cloud service provider currently offering big language models, but it also makes some of the features of Microsoft Cloud Services overlap with those of OpenAI's products.

This overlap means that Microsoft and OpenAI sometimes have to "compete" for customers.

An internal Microsoft document instructs Azure salespeople to tell potential customers that "OpenAI offers limited enterprise-class functionality and security and privacy services," while "Auzre OpenAI's services not only offer the same functionality as OpenAI, but also address compliance, data privacy, and security needs. This rhetoric did help Microsoft cloud services attract customers, such as marketing software startup Touchcast, which chose to access the related Big Model service through Azure because of the security and performance guarantees.

OpenAI, on the other hand, is also looking to connect directly with big customers, including Microsoft's rival Salesforce, while at the same time creating a "time lag" by delaying granting Microsoft access to Big Model in order to work with customers. For example, in March, after OpenAI had signed up Snap and Instacart, Microsoft Cloud Services announced a preview of the ChatGPT feature a week later, and after the GPT-4 language model was released, Microsoft Cloud Services was granted access to GPT-4 only after companies like Duolingo and Stripe had already paid OpenAI directly. 4 access.

Currently, many B-side customers have to choose between "contracting directly with OpenAI" and "using the big language model through Azure," with the latter generally preferred by customers who are already Azure users.

It's almost blatantly clear that not only is there a conflict of interest between Microsoft and OpenAI, but that both sides have acted accordingly.

Some people may wonder why Microsoft is so eager to make money by selling APIs for AI big models when it has just received $10 billion from Microsoft and can now do research without worrying.

From the source, OpenAI's strategic goal is to achieve AGI (General Artificial Intelligence) as soon as possible, and the way to achieve this goal is to use a lot of computation to continuously train, optimize, and iterate on the existing big models. This means that OpenAI faces high computational and R&D costs in implementing AGI.

The MIT Research Review reported that "OpenAI is under increasing pressure to commercialize, and founder Altman sees the need to make money to continue his research. So even though OpenAI started out as a purely non-profit organization with a vision of never allowing AI technology to be monopolized by commercial companies, in 2019 OpenAI made a structural change - adding a for-profit entity with a "profit cap" (a return to investors of 100 times or less). OpenAI LP is a for-profit entity with a "profit cap" (100x return to investors).

Some industry insiders believe that this is just a small conflict between the sales departments of the two companies, so there is no need to make a fuss; however, others believe that the two are currently in the honeymoon period of cooperation, so these small conflicts will not affect too much, but the overlap between the two businesses is always a hidden problem. At the same time, from a longer-term perspective, the two companies may inevitably fight for users in the C-tier products in the future.

ChatGPT's user growth is obvious to all. According to Similar Web's data at the beginning of this year, there are about 13 million unique visitors using ChatGPT every day, with a growth rate of more than 100% year-on-year. In March this year, OpenAI even announced the release of a plug-in feature that allows people to call third-party software directly in the dialog box when using ChatGPT, indicating that as the number of visitors to ChatGPT increases, OpenAI is trying to build its own new application ecology through ChatGPT: using ChatGPT as a platform and vertical AI applications as an extension to help users intelligently use ChatGPT. With ChatGPT as the platform and vertical AI applications as the extension, OpenAI is trying to help users to do various things intelligently, which will have a disruptive impact on the future C-application ecology.

After the announcement of the plug-in, the number of unique visits to OpenAI's website soared to 1 billion in March. According to research by marketing agency VezaDigital, OpenAI climbed nine spots in just one month to become the 18th most visited website in the world in terms of online visits alone.

In contrast, although Microsoft said at this year's developer conference that its main products will integrate generative AI features and enhance productivity through Copilot, the problem is that once OpenAI's ChatGPT becomes a super portal for C-tier applications in the future, Microsoft's products will be relegated to a part of the entire ecosystem rather than the main character.

In other words, although Microsoft and OpenAI are now partners and can cooperate with each other in terms of technology, in the future AGI era, if OpenAI has the ambition and goal to build a new underlying ecology through ChatGPT, then the position of Microsoft Windows system will be threatened. At that time, there is still a strong uncertainty as to who will be the most basic and important portal.