Big Tech AI Valuations, Circular Finance and Chinese Model Shift

In the last five years, big tech companies have experienced unprecedented valuations, far exceeding historical norms. While conventionally this growth would be celebrated as a sign of a prosperous economy, the Bank of International Settlements issued a stark warning in late June 2026: AI spending and circular financing have become so uncontrolled that they pose a significant risk to the global economy. Big tech, once an economic driver, is now seen as a potential threat.

The Rise of Chinese AI Models and Shifting Landscape

A new factor fueling this concern is the rapid advancement of Chinese AI models. These models have drastically closed the performance gap with their Western counterparts, becoming "good enough" for most AI applications while being significantly cheaper. They offer the advantages of offline use, 100% data control, and customizability. American companies are taking notice and switching to these open-source and open-weight models. Major players like Cursor, Coinbase, Shopify, Airbnb, Uber Eats, Siemens, and even Microsoft are making this transition. This shift raises questions about the trillions of dollars invested in AI if the anticipated revenue is already diminishing.

The Mispricing of LLMs and Real-World Challenges

Despite the long-term revolutionary potential of AI, particularly in coding and healthcare, the industry appears to have mispriced Large Language Models (LLMs) due to excessive hype. Beyond recent data center project collapses, LLMs face significant real-world problems. A survey of nearly 2,500 companies revealed that for every dollar spent on AI, only 18 cents translates into production. The remaining 82 cents is spent on fixing AI-generated bugs, reworking AI outputs, or addressing general friction. Such a poor return on investment would typically lead to a technology being discarded, yet investment continues to pour in. Even top AI CEOs are moderating their language; Sam Altman, for instance, expressed delight at being wrong about an AI-driven job apocalypse. Some analysts fear that the market has moved beyond a bubble, indicating a potentially broken economy.

Big Tech's Strange Behavior and Artificial Valuations

Compared to the 2010s, when tech investment was relatively slow, the current era is marked by immense capital flow. While Apple's iPhone, Amazon's logistics, and early Google AI research (like identifying cats from toasters) were notable, the scale of today's tech valuations is unprecedented. The six largest tech companies (Amazon, Apple, Alphabet, Microsoft, Meta, and Nvidia) saw their combined revenue exceed a trillion by the end of 2020, their valuation reached nearly $8 trillion, and by the end of 2025, it had exploded to over $20 trillion, now exceeding $23 trillion. This growth is so monumental that some believe the bullish tech sector is the only thing preventing a massive recession in the American economy.

However, if this tech boom is a bubble, its behavior differs from past bubbles. A significant factor is AI and data center construction, which have meaningfully contributed to global GDP. Harvard's John Ferman notes that information processing systems and software (data centers) accounted for 92% of the increase in US GDP in the first half of the year, suggesting that GDP might be a broken indicator of economic prosperity when much of this "production and consumption" is driven by the same companies reaching unprecedented valuations. This ties into the concept of circular financing.

Circular Financing and Accounting Tricks

Forensic accountants John While and Kevin Kohhaki highlight how major companies like Microsoft and Google, despite appearing profitable, are engaging in financial engineering. They explain that while initial net income and cash flow look impressive, significant capital investments in data centers, stock-based compensation, and related buybacks drastically reduce actual earnings and free cash flow. Companies that were "printing cash" five years ago are now seeing their free cash flow "fall off a cliff" due to massive spending on data centers.

A prime example of this circular financing is the relationship between Google and Anthropic. Google's first-quarter 2026 profits showed an 82% increase, attributed to its cloud unit. However, Google was simultaneously laying off staff in its cloud division to "reinvest in growth areas such as AI." The profit jump was attributed to "other income." Further investigation reveals that Google has been investing tens of billions into Anthropic. One month after Google's investments, Anthropic committed to spending $200 billion with Google's cloud over the next five years. This pattern, where Google invests in Anthropic, and Anthropic then spends that money on Google's cloud services, creates artificial revenue and inflated valuations. Sasha Yanshin, a financial product expert, describes this as a cycle where Google invests, receives the money back as revenue, and then reinvests at a higher valuation, creating massive profits without actual capital changing hands. A similar pattern is observed with Amazon.

Disappointing AI Performance and Loss of Faith

Despite the massive investments, the return on investment for AI is often lacking.

• Starbucks: Dropped an AI-powered inventory tool after it proved to be a "catastrophic failure," misidentifying and mislabeling items, requiring staff to manually check everything.
• Duolingo: Its CEO promoted AI-assisted learning with plans to cut staff, but the AI-generated output was so poor that the company had to reverse course and appeal to consumers.
• Microsoft CEO Satya Nadella: Stated that LLM models alone are insufficient and unstable, emphasizing the need for a proper ecosystem where humans provide ideas and judgment, with AI tokens only used for building.
• Loss of faith in total AI replacement: Megan Slinsky of Robert Half notes that organizations are reassessing expectations, recognizing AI's limitations. Box CEO Aaron Levy warns that AI fanaticism has led to major losses, stressing the need for human oversight in software development and maintenance.

This situation has left a bitter taste, as people were fired for unproven technology, while big tech companies seemingly paid each other to artificially inflate valuations.

Nvidia, SpaceX, and Alleged Backroom Dealings

One of the most significant examples of questionable practices involves Nvidia and SpaceX. Michael Burry, known for predicting the 2008 crisis, expanded his bearish stance on Nvidia in May 2026, highlighting a complex web connecting pension funds, insurers (Athene), private credit firms (Apollo), and the AI infrastructure between Nvidia and Elon Musk's XAI.

A key deal involves Nvidia's sale of $5.4 billion worth of its advanced GB200 GPUs to a company called Valor, purportedly for buying and leasing data center infrastructure, primarily for XAI. Skeptics, including Burry, allege that these chips are technically owned and leased by Nvidia through a not-so-separate company (Valor) to legitimize the transaction on paper. This arrangement allows Nvidia to claim $5.4 billion in sales revenue, XAI to power its Grock AI, and Apollo to collect fees, all while retirees unknowingly finance the operation through their pensions via Athene.

This context sheds new light on SpaceX's IPO, which is floating around the $2 trillion mark, especially after its $250 million acquisition of XAI. SpaceX's IPO filing is under "computer programming and data processing," with 85% of its proposed market being AI. SpaceX's subsequent purchase of the AI coding company Cursor further solidifies this shift.

While critics like Michael Burry might be labeled as conspiracy theorists, these transactions are often conducted openly. Although the SEC appears to be inactive, these actions are not explicitly illegal. It seems big tech has found ways to "have its cake and eat ours," fabricating demand, providing exit liquidity to wealthy investors, and having ordinary people foot the bill. This suggests that big tech might not be preparing for further growth but rather seeking insurance against future uncertainties.

The Importance of Critical Thinking in an AI-Driven World

In this era of technological convenience, the ability to think critically and solve problems is more crucial than ever. Brilliant offers a personal tutor for math and coding that guides users through problems, adapts to their reasoning, and fosters intuition rather than passive information consumption. With courses ranging from basic math to calculus, coding, and debugging, designed by experts from institutions like MIT and Harvard, Brilliant provides an engaging and flexible learning experience.