The AI Deals Enriching Silicon Valley’s Tech Giants

Summary
Microsoft, Amazon and Google are getting billions in revenue from providing expensive cloud services to the startups they are backing.Amazon, Google and Microsoft have spent the past year investing billions of dollars in artificial-intelligence startups—while also charging those fledgling companies a similar amount to use their cloud platforms.
The deals are making the big tech firms both the largest backers and most direct beneficiaries of these startups, reflecting how some of the AI boom’s biggest rewards keep going to the most powerful players. The value of the tech giants’ stakes could shoot up if the startups take off. And if not, they still will have turned chunks of cash into revenue.
For the startups, the deals give them the cash they need to train advanced AI models as well as access to the scarce computing power essential for developing and deploying products such as ChatGPT.
The symbiosis between AI startups and the tech giants is sidelining the venture capitalists that typically back young companies. Those investors rarely spend in the billions and have been more wary of paying the high valuations that come alongside the large deals.
In September, Amazon announced a deal to invest up to $4 billion into Anthropic, a rival to ChatGPT creator OpenAI. Not part of the announcement was another deal the Seattle giant struck: Anthropic committed to spend $4 billion on Amazon’s cloud platform, Amazon Web Services, over the next five years, according to people familiar with the matter.
Google, which had invested in Anthropic early this year, recently agreed to pour up to $2 billion more into the startup, The Wall Street Journal reported last week. That new commitment came months after the startup agreed to spend more than $3 billion on Google Cloud.
Both companies followed Microsoft, which has invested $13 billion into OpenAI. Meanwhile, OpenAI is spending billions of dollars on Microsoft’s cloud.
These types of deals have made the tech giants by far the largest backers of these ambitious and costly AI startups. Microsoft, Google and Amazon have poured nearly $20 billion into Anthropic and OpenAI alone, with well over half that amount coming this year. Given that these startups’ biggest cost is on cloud computing, the majority of that cash will likely come back to these investors in the form of cloud revenue.
“For the cloud providers, it’s a masterstroke," said Margaret Jennings, the co-founder of the AI startup Kindo and a former OpenAI employee. “Their investment money gets paid back, and now they have direct access to how these startups’ research and product teams are building out their strategy that they can help influence and steer."

Since the viral success of ChatGPT a year ago, a group of startups have been racing to deliver the next breakthrough in generative AI, the technology enabling software to converse like humans. Building such technology requires supercomputers outfitted with high-end chips and billions of dollars to run them. Something the tech giants have—and nearly no one else does.
Microsoft pioneered this type of partnership for generative AI startups four years ago when it invested $1 billion in OpenAI. In exchange, OpenAI agreed to exclusively train its software on Azure’s servers and release its products through the platform. Microsoft followed up at a much larger scale in January when it announced plans to invest $10 billion.
As customers have signed up to use ChatGPT and build tools on its underlying technology, Microsoft has reaped the benefit. In the most recent quarter, Azure revenue grew at a robust 29% compared to the quarter a year earlier; Microsoft executives said 3 percentage points of that growth came from AI spending alone. That translates to around $400 million in AI spending on Azure, based on estimates of the businesses’ revenue. The majority of that spending is coming from OpenAI and products built on its software, analysts say.
Venture capitalists who compete with the cloud giants have noted the circular nature of these deals.
“Using your balance sheet to potentially artificially inflate revenue is an area of concern for auditors," said Bill Gurley, a venture capitalist at Benchmark who invested early in Uber Technologies. “It’s something that deserves scrutiny."
This type of deal is permissible as long as it’s driven by a legitimate business purpose and not just to inflate revenue, said Christopher Armstrong, an accounting professor at Stanford University’s business school.
The cloud giants have made similar strategic investments in the past, but none at the scale of these recent deals.
Executives at the tech companies say these investments and cloud commitments are struck by different teams. They say their aim isn’t only to bankroll a big-spending customer but to make a profit on their investments.
“The investments have to stand on their own merits, full stop," said Kevin Ichhpurani, corporate vice president at Google Cloud, during an interview in August. He said Google’s investments and cloud contracts are separate agreements.
In addition to its deals with Anthropic, Google has struck a string of similar investments and cloud contracts with a host of smaller startups, investors involved in the deals say.
OpenAI’s deal with Microsoft stands out from the pack in that it is the only one that also came with exclusivity—all of its cloud spending is on Azure. Other companies such as Anthropic run on multiple cloud providers such as AWS and Google.
Google hasn’t reported the same kind of AI-driven bump in cloud revenue that Microsoft did. Shares in Google parent Alphabet slipped more than 10% in trading last week after the company reported slower-than-expected growth in the cloud business. Google Cloud executives have touted the unit’s close relationship with valuable AI startups, betting they will become even bigger customers over time.
No one knows whether today’s big AI startups, which all currently lose money, will become enduring businesses. But at least for now, their valuations keep shooting higher. OpenAI recently began a process of selling employee shares at a price that would value the company at more than $80 billion—more than doubling where it was earlier this year.
Buying in at these sky-high valuations is a problem for a venture-capital firm that is trying to score a big return. It is a lesser issue for a tech company that is mostly trying to support an ecosystem of startups, said John Somorjai, who leads investments for Salesforce.
Salesforce isn’t a cloud-computing provider, but it has been among the most active corporate investors in the space. The seller of customer relations management software recently won a deal to lead an investment in Hugging Face, a site that hosts AI software, that valued the company at $4.5 billion.
Somorjai said Salesforce beat out offers from multiple venture-capital firms. He said Hugging Face was looking for a backer that could also give it access to new customers. And Salesforce was excited by the opportunity to connect their customers to Hugging Face’s catalog of AI software.
“It’s wonderful when you have a big return, but it’s not the reason we would make an investment," Somorjai said.
Miles Kruppa contributed to this article.
Write to Berber Jin at berber.jin@wsj.com and Tom Dotan at tom.dotan@wsj.com