Behind the scenes of ChatGPT: Who is the winner and who is the fool in the "complex deal" between Microsoft and OpenAI?

　　In fact, OpenAI's approach is to lease the company to Microsoft, and the lease period depends on OpenAI's profitability.

Intuitively, the terms of the investment are extremely favorable for Microsoft.

　　In the end, judging who is the biggest winner in this deal will depend a lot on how valuable ChatGPT and other generative AI technologies OpenAI currently has in its portfolio (such as the text-to-image generator DALL-E 2) , and how close OpenAI is to achieving its stated mission of artificial general intelligence (AGI).

　　After the chat robot ChatGPT became popular, news came out in January that Microsoft was negotiating to invest $10 billion in OpenAI, the development agency of ChatGPT, making the company's valuation close to $29 billion.

According to the currently known information, this is a complicated transaction, which contains many details worth thinking about, as well as many unknown transaction methods.

　　"Fortune" reporter Jeremy Kahn (Jeremy Kahn) disclosed the details of the transaction based on sources familiar with the transaction and some internal documents.

"There's an old saying in business, if you don't know who's the fool in a deal, it's you," Kahn said.

Who's the fool in the Microsoft-OpenAI deal?

　　unusual investment structure

　　In the first phase, Microsoft is entitled to 75 percent of OpenAI's profits until it recoups the $10 billion, plus an additional $3 billion Microsoft has already invested in the company (2019), according to sources familiar with the deal and internal documents. The $1 billion publicly invested in the year, and another $2 billion that Microsoft invested in OpenAI in a low-key manner that year).

　　In the second stage, when OpenAI's profit reached $92 billion, Microsoft's share of OpenAI's profit dropped to 49%.

Meanwhile, other venture investors and OpenAI employees will be entitled to 49 percent of the company's profits.

　　In the third phase, after OpenAI's profits reach $150 billion, Microsoft and other venture investors' shares in OpenAI will be returned to OpenAI's non-profit foundation.

　　One piece of information needs to be added here. The current OpenAI is composed of the for-profit company OpenAI LP and the non-profit parent company OpenAI Inc.

In 2015, OpenAI was established as a non-profit organization, the goal is to achieve general artificial intelligence (AGI) in a safe way, so that all human beings can benefit equally, rather than create profits for the company's shareholders, so founder Sam Altman ( Sam Altman holds no shares.

But in 2019, OpenAI transformed from a non-profit organization to a for-profit organization with a "profit cap", and established a unique structure - there is a cap on the return on investment (investment profit does not exceed 100 times the investment amount), and any excess All will be returned to non-profit organizations.

　　In fact, OpenAI's approach is to lease the company to Microsoft, and the lease period depends on OpenAI's profitability.

According to Kahn, this investment structure is very unusual.

　　There are still many details about the deal that are not known: How much cash will Microsoft actually pay OpenAI, for example?

　　Microsoft said that as part of the partnership, it is investing more in building artificial intelligence supercomputing clusters in its Azure (Microsoft cloud service) data centers.

Most of what OpenAI actually gets from the deal is probably the right to use these supercomputing clusters at little to no cost, and the $10 billion is mostly an "in-kind payment" for computing resources.

　　It is also unknown how many years Microsoft will deliver the $10 billion to OpenAI.

Additionally, it's unclear whether Microsoft would have to pay OpenAI anything in terms of licensing or royalties to use OpenAI's technology in its product suite.

If Microsoft integrated ChatGPT into Bing, would OpenAI see a small gain in every search?

　　What benefits will Microsoft get?

　　For such a transaction, who will benefit more from it?

　　Intuitively, these terms are clearly extremely beneficial to Microsoft.

Even if it paid out $10 billion all in cash for the year (which Microsoft seems to say isn't the case), that's only 15% of Microsoft's trailing 12-month free cash flow of $63 billion.

That's not a lot of money for a technology that could give Microsoft a host of advantages, especially since this is Microsoft's first real attempt to dent Google's dominance in search.

　　Search generates $150 billion in revenue for Google, according to a rough calculation by Fortune.

By 2021, Google will have about 90% of the global search market, while Microsoft's Bing will have only 3%.

If adding ChatGPT to Bing allowed Microsoft to increase its market share to 10%, the extra revenue might have compensated Microsoft for its $10 billion investment in OpenAI.

　　At the same time, here's an important consideration: It's not clear that a company can generate nearly as much revenue from a chat-based search interface as it does from search.

Because returning a single, coherent answer to a question greatly reduces the likelihood that people will click on any links, such as the ads Google displays next to its search results.

　　Microsoft has many other benefits, such as being able to provide OpenAI's models to Microsoft Cloud customers, making Microsoft Cloud more attractive.

Wall Street analysts estimate that the Microsoft cloud brought in about $35 billion in revenue for Microsoft last year (Microsoft does not break out its cloud revenue or profits in its financial statements).

So even a small boost could lead to about $1 billion in revenue boost.

Microsoft, then, would have even more potential advantage by incorporating OpenAI's technology into everything from its Office software suite to its Xbox game console.

　　Still, some of the benefits Microsoft is currently reaping could be quickly eroded by competition.

Google has launched its own chatbot, Bud, and Google's sister company, DeepMind, has a chatbot called Sparrow.

At the same time, many other companies are building their own large-scale language models (the technology on which ChatGPT is based), so Microsoft may not gain an advantage for long.

Still, $13 billion is a reasonable amount for a company of Microsoft's size and wealth to pay for the aforementioned potential advantages.

　　What's more, Microsoft will essentially own OpenAI, which could be for a long time since it's entitled to a share of OpenAI's profits until OpenAI hands over $105 billion to Microsoft (according to a deal outline seen by Fortune, which is Capped profits to which Microsoft is entitled to the $13 billion it has invested thus far.)

　　What price will OpenAI pay?

　　According to public information, OpenAI is expected to lose $544 million last year.

While OpenAI also expects its revenue to increase significantly, from less than $30 million last year to more than $1 billion in 2024, there's no clear indication of when OpenAI will turn a profit, or how profitable it might be.

It's worth noting that even Apple, a super-profitable company, now only has about $100 billion in annual net income.

　　Assuming OpenAI can turn a profit in 2024 with a net profit margin of 35%, its net income will reach $350 million, and Microsoft will get 75% of it, or $262 million.

Even assuming OpenAI can double its net profit every year (a very optimistic assumption), it will take another decade for Microsoft to reap the rewards, meaning OpenAI can become independent again.

　　What will it cost OpenAI to be largely owned by Microsoft for most of the next decade?

　　Even Microsoft's initial investment in 2019 pushed OpenAI to focus more on creating commercial products, former OpenAI employees told Fortune.

It helped solidify the focus on large language models (LLMs), limiting other avenues of research.

If LLM is not the path to AGI (artificial general intelligence), OpenAI may well be distracted from building for Microsoft and miss the next big AI breakthrough.

Microsoft may not really care whether OpenAI can achieve AGI, as long as OpenAI's products are commercially useful to Microsoft.

But for OpenAI, this outcome will be a failure.

　　If OpenAI can use Microsoft's supercomputing infrastructure to actually achieve its AGI goals and stay ahead of any competitors, then it ends up being a bad deal for Microsoft.

In that case, the benefits of the technology are so great that OpenAI might actually be able to pay Microsoft a quick $105 billion profit on its $13 billion investment, but for AGI, possibly one of the most powerful technological advances in human history, has no ownership rights whatsoever.

Interestingly, it could also be a bad deal for Microsoft if one of OpenAI's competitors (perhaps less focused on launching a commercial product) gets AGI first.

　　So judging who is the biggest winner in this $10 billion deal will largely depend on the value of ChatGPT and OpenAI's other generative AI technologies currently in its portfolio (such as the text-to-image generator DALL-E 2). How big, and how close OpenAI is to fulfilling its stated mission of artificial general intelligence (AGI).