Q1 2026 closed with record-breaking profit figures from America’s tech giants. Yet behind the market’s optimism lies a financial picture full of paradoxes. In just the first three months of the year, the four largest tech conglomerates burned through $130.65 billion in capital expenditure (CapEx) to build AI infrastructure.
Amid that massive wave of investment, financial reports still showed strong profit growth. Alphabet reported a sharp rise in earnings, Amazon continued recording its best quarters in history, and the market kept betting that AI would unlock the next major growth cycle for the tech industry. But what is truly propping up those profit figures, as infrastructure-building costs continue to balloon?
That is the question financial analytics platform Bull Theory recently raised in an article that sparked heated debate on social media. Combined with investigative reports from Fortune and independent experts, analysts argue that the future $2 trillion cloud revenue ecosystem may be shaped by a sophisticated accounting loop engineered by Big Tech itself.
How the cloud revenue loop operates
The core of analysts’ skeptical argument lies in how Big Tech structures its financial relationships with AI startups. When Microsoft, Amazon, or Alphabet invests billions into OpenAI or Anthropic, a significant portion of that investment does not exist as ordinary cash, but rather as cloud credits or commitments to use computing infrastructure.
According to an FTC report cited by ProMarket, AI startups then use the investor’s own infrastructure to train and operate AI models. That spending is subsequently recorded as cloud revenue for Big Tech. This is also why many analysts have begun paying closer attention to the increasingly tight financial relationship between AI infrastructure providers and AI startups.
Analysts estimate OpenAI’s annual cloud bill has now reached approximately $60 billion, significantly higher than the company’s current revenue. Meanwhile, data compiled from FTC reports shows Anthropic spent around $2.66 billion on Amazon Web Services in just the first nine months of 2025. By Bull Theory’s calculations, that figure is equivalent to 100% of the revenue the AI startup generated in the same period.
The chart above illustrates the growing dependence of major cloud computing players on the AI wave. According to data from The Information, OpenAI and Anthropic alone now account for a very large share of the future revenue backlog of cloud providers such as Microsoft, Oracle, Google, and Amazon.
For Microsoft and Oracle, more than half of that backlog is believed to come directly from AI spending commitments. This reflects the enormous infrastructure consumption of AI startups, but also raises concerns that current cloud growth is overly reliant on a handful of “AI whales” rather than the diverse enterprise customer base of the past.
This model draws comparisons to the dot-com era of the early 2000s, when telecom companies like Global Crossing and Qwest Communications engaged in infrastructure swap deals to book revenue. However, unlike many controversial transactions of the dot-com era, today’s AI arrangements remain fully legal and compliant with current accounting standards.
When AI startups generate both revenue and profit for Big Tech
If cloud revenue is the first layer of this loop, the more significant layer lies in how Big Tech recognizes profit from its AI investments.
Under the current U.S. accounting standard ASU 2016-01, companies must periodically update the value of equity investments to reflect the latest market valuations. This means that when Anthropic or OpenAI is valued higher in a new funding round, companies holding equity stakes — such as Amazon or Alphabet — can recognize that increase as unrealized gains.
What has caught the financial community’s attention is that the same investment can now produce two parallel effects. The AI startup uses cloud infrastructure to generate revenue for Big Tech, while the startup’s rising valuation allows Big Tech to book additional accounting profits.
Fortune quoted Robert Willens, a tax and accounting expert who previously taught at Columbia Business School, as saying that this accounting treatment itself is not unusual, since companies are required to update investment values when startups raise funds at new valuations. However, he noted that what has drawn market attention is the very mechanism by which AI startup valuations are formed.
For public companies like Apple, share value is determined by the open market with millions of investors buying and selling every day. But for private startups like Anthropic, valuations are primarily established through funding rounds involving a small group of strategic investors — with Amazon and Alphabet themselves among the most significant.
According to Willens, when Big Tech continues to inject more capital or expand infrastructure commitments to AI startups, it also helps reinforce the value of the very assets they hold. As valuation increases, the difference continues to be recorded as accounting profit.
It is precisely the Q1 2026 financial reports that have drawn attention to this argument. In Alphabet’s financial report, the company recorded approximately $28.7 billion in profit from revaluing equity investments in private companies, largely related to Anthropic. That figure accounted for nearly half of Alphabet’s record total profit of $62.6 billion announced for the quarter.
Amazon also disclosed a similar gain. In its quarterly financial report, the company stated that net income included approximately $16.8 billion in pre-tax gains from its investment in Anthropic, representing more than half of the quarter’s total pre-tax profit.
In response to Fortune, Amazon said the increase in value came from Anthropic’s Series G funding round and the conversion of a portion into preferred shares. The company also revealed that its $8 billion investment in Anthropic is now valued at over $70 billion, while emphasizing that this investment is separate from the commercial relationship between the two parties.
Meanwhile, the impact on actual cash flow remains stark. Despite reporting record profits, Amazon’s free cash flow in the same period declined sharply as the company spent over $44 billion in cash to expand data centers and AI infrastructure — a metric that reflects how much cash a business actually retains after deducting operating costs and capital expenditures.
It is precisely this structure that has begun to concern many analysts.
As AI startup valuations continue to rise, Big Tech can book additional unrealized gains, thereby reinforcing growth expectations in the stock market. This also triggers a chain reaction affecting index funds and pension funds. As tech company market caps continue to grow, their weighting in indices like the S&P 500 also increases. ETFs, index funds, and many 401(k) retirement funds therefore automatically allocate more capital to tech stocks.
This is why many believe AI is no longer just a technology race, but also a financial structure where investment, cloud revenue, accounting profit, and stock prices continuously influence one another. If AI startup growth slows or investment capital dries up, the unrealized gains currently on the books could quickly reverse.
That is also why a growing number of people argue that AI is no longer just a technology race, but the modern version of “taking from the left pocket and putting into the right” — where Big Tech simultaneously invests, generates revenue, and books profit for itself at a scale large enough to move Wall Street.
Not everyone believes in the “AI bubble”
While the financial structure that Bull Theory has identified continues to generate controversy, the market also has no shortage of counterarguments. Unlike many dot-com era internet companies that lacked clear business models, today’s AI platforms are generating real products and real revenue.
APIs from OpenAI and Anthropic have already attracted a large number of paying enterprise customers, while AI chatbots are gradually becoming mainstream tools for general consumers. This suggests that AI does not exist purely on financial expectations or startup valuations.
Additionally, many investors argue that Big Tech’s aggressive AI infrastructure buildout is not solely serving language models. The data center systems of Amazon, Microsoft, and Alphabet currently run the majority of the global internet while serving millions of businesses outside the AI sector. This wave of investment is also driving strong growth across the entire technology supply chain, from GPUs and HBM memory to energy, cooling systems, and data center construction.
For companies with massive cash flows like Meta or Alphabet, AI is now viewed as a long-term investment to protect their position in the next technology cycle. While the commercial effectiveness of AI remains a question mark in the short term, many investors believe Big Tech still has ample resources to sustain its infrastructure expansion pace for years to come.
Will the AI boom become a new revolution or a new bubble?
AI is clearly driving real changes across the technology industry, but the way financial markets are pricing this race is also fueling considerable debate. Analysis from Bull Theory, together with investigative pieces from Fortune and ProMarket, has highlighted the increasingly tight link between AI investment, cloud revenue, and Big Tech’s accounting profits.
However, that does not confirm that the entire current AI boom constitutes “fake revenue.” The biggest issue is that the pace of infrastructure investment is growing significantly faster than the industry’s ability to commercialize AI. If actual revenue continues to grow strongly, current investments may be viewed as necessary groundwork for a new technology platform. But if growth slows, the expectations currently priced into Big Tech stock valuations could also face significant pressure.
That is why a growing number of experts are calling on regulators and investors to more closely scrutinize how Big Tech recognizes profit from AI investments, and to clearly distinguish between core operational growth and unrealized gains from startup valuations.