Everyone is learning from Microsoft and Nvidia! The so-called "bold bet on AI" by tech giants: investing to generate revenue and gain a competitive edge.

For tech giants, investing in AI startups seems like a "sure bet": they invest billions of dollars in AI startups, but at the same time, they also generate revenue through cloud platform usage fees or GPU sales from these companies.

Last week, it was reported that Google has agreed to invest another $2 billion in AI startup Anthropic, which had just agreed a few months ago to pay over $3 billion to Google Cloud for its services.

Anthropic is one of OpenAI's biggest competitors. In September of this year, Amazon also invested a staggering $4 billion in Anthropic. According to insiders, the two parties have also reached another agreement: Anthropic has committed to paying $4 billion in usage fees to Amazon Web Services (AWS) over the next five years.

According to statistics, Microsoft, Microsoft, and Amazon have invested nearly $20 billion in Anthropic and OpenAI alone, with over half of the funds being invested this year.

For AI startups, their biggest cost is cloud computing, which happens to be owned by tech giants. Therefore, most of the cash investments they receive may be returned to these tech giant investors in the form of cloud service revenue.

This means that regardless of how these startups perform, tech giants are not only their biggest supporters but also their biggest beneficiaries. If the startups take off, the tech giants' holdings will skyrocket; even if they don't, the giants can still generate revenue from their cloud platforms.

For example, as more and more customers use ChatGPT, Microsoft has already benefited from it.

Microsoft's recent earnings report showed that in the latest quarter, revenue from its cloud service Azure grew strongly by 29% YoY. Microsoft executives stated that 3 percentage points of this growth came from AI spending, which means that AI revenue on Azure is approximately $400 million. Analysts told the media that most of this revenue comes from OpenAI and products developed based on its software.

It should be noted that the deal between OpenAI and Microsoft is the only exclusive deal: all of OpenAI's cloud spending is on Azure, while other startups like Anthropic operate on multiple cloud providers.

Margaret Jennings, co-founder of AI startup Kindo and former OpenAI employee, said:

This is a masterpiece for cloud providers.

Their investments have paid off, and now they can directly understand how these startup companies' research and product teams formulate strategies. They can help influence and guide these strategies.

Circular Transactions

Venture capital firms have already noticed this circular transaction between investment giants and startups. Benchmark's venture capitalist Bill Gurley warns that this approach may carry the risk of inflating revenue:

Using your balance sheet, you may artificially inflate revenue, which is an area of concern for auditors. This matter deserves careful scrutiny.

Executives of tech giants have stated that these investments and cloud computing are handled by different teams, and their goal is not only to provide funding to customers but also to profit from their investments.

Kevin Ichpurani, Vice President of Google Cloud, stated in an interview with the media in August that Google's investments and cloud contracts are separate agreements.

Investments must be based on their own value, and that's it.

However, Google did not experience significant growth in cloud revenue from artificial intelligence like Microsoft did. After Google's parent company, Alphabet, announced lower-than-expected growth in its cloud business last week, its stock price fell more than 10% in trading.

Investors involved in the transactions have stated that in addition to the deal with Anthropic, Google has reached similar investment and cloud computing contracts with many smaller startups.

Christopher Armstrong, an accounting professor at Stanford Graduate School of Business, believes that as long as these transactions are for legitimate business purposes and not just to increase revenue, they are allowed.

Microsoft's Model: Dual Binding

As one of the pioneers in this wave of AI frenzy, Microsoft is one of the earliest tech giants to adopt this circular transaction model. Through these transactions, Microsoft not only gained substantial returns but also took a commanding lead in the "AI race" among the giants.

In 2019, when OpenAI was facing its most challenging times, Microsoft invested $1 billion in OpenAI to provide the powerful server infrastructure needed for developing large-scale models.

Over the course of five years, Microsoft has invested a total of $13 billion in OpenAI. In return, Microsoft received a 49% stake in OpenAI, a share of future profits, and the right to use its technology.

Microsoft also integrated OpenAI's technology into its search engine Bing, marketing software, GitHub coding tools, and Microsoft 365 office software, creating a comprehensive and powerful AI service system for individuals, businesses, and developers.

The collaboration with OpenAI helped Microsoft seize the initiative in the generative AI field, allowing this traditional tech giant to regain "high growth, high gross profit, high market value," and experience another shining moment despite lagging behind Google in the AI field.

According to Michael Turrin of Credit Suisse, this series of integrations will increase Microsoft's revenue by $30 billion annually, with about half of it coming from Azure.

However, despite OpenAI's subsequent escape from survival difficulties and rapid growth, its ability to generate revenue is severely constrained due to being "deeply bound" by Microsoft in terms of equity and technology. This most valuable startup is gradually being eroded by Microsoft. As mentioned earlier, Microsoft has obtained the right to use OpenAI's technology and is bundling it with its other products to sell downstream, directly squeezing OpenAI's revenue. Not only that, Microsoft not only "intercepts" most of the sales revenue from OpenAI on Azure, but also shares most of OpenAI's profits.

In addition, OpenAI is not the exclusive supplier of large models for Microsoft. Microsoft Azure also provides large models from Meta and Hugging Face, profiting from both ends.

NVIDIA's Model: GPUs are the "Hard Currency"

The partnership between Microsoft and OpenAI involves profit and technology sharing, while NVIDIA, which holds the "hard currency" of AI, GPUs, has a more subtle model.

Wall Street News has previously introduced that Coreweave, a cloud service provider invested by NVIDIA, has purchased a large number of NVIDIA GPUs and leased them to AI and machine learning development companies.

Since the beginning of this year, CoreWeave has raised more than $2.7 billion through debt and equity financing to invest in more chips and data centers. NVIDIA has also included CoreWeave in the first batch of shipments of its advanced AI chip, H100.

However, due to NVIDIA's support for CoreWeave, in the case of a shortage of H100, a large number of new cards were still allocated to CoreWeave, and NVIDIA directly participated in the investment, which led to rumors of "NVIDIA's deception" in the market.

There are also more cloud service providers who borrow money to buy NVIDIA chips and use the money they earn to repay the loans.

Crusoe Energ, which was also a cryptocurrency "miner," joined the ranks of cloud service providers this year and leased servers with H100 GPUs to startups.

It is reported that in order to purchase a large number of scarce NVIDIA H100 chips, Crusoe borrowed $200 million in financing.

What's special is that after purchasing the chips, Crusoe used them as "collateral assets" for loans. It will take Crusoe three and a half years to repay the loan, including interest. However, Crusoe believes that these chips can bring returns higher than their $32,000 price tag and can continue to generate profits for the company for seven years.

According to Bill Libby, CEO of Upper90, NVIDIA GPUs are "a new category of debt assets." Before signing the debt agreement, he must consider how to value the chips and how to adjust the valuation if the chip version is updated. This means that the fate of these cloud service providers is tied to NVIDIA. As long as NVIDIA's chips continue to dominate the market, the cloud service providers will be profitable.