Key Moments
One Chinese AI Model Wiped Out $1 Trillion In A Single Day — And They're Just Getting Started
Impact TheoryWant to know something specific about what's covered?
We've already dissected every moment. Ask and we will deliver (with timestamps).
Key Moments
Cheap Chinese AI models trained on stolen US tech could collapse the US AI market and trigger an economic recession due to massive debt burdens.
Key Insights
80% of the US stock market's gains over the last three years have come from AI, with the 10 largest S&P 500 companies making up over 40% of the index.
China is allegedly using a technique called 'distillation' to train cheaper AI models on outputs from expensive US models, potentially exceeding 16 million queries via fake accounts, according to Anthropic.
The MIT found that 95% of corporate AI projects produced no measurable impact on profits, despite the immense cost of AI infrastructure development.
Oracle is spending 57% of its revenue on AI buildout, Microsoft is around 45%, and overall, AI infrastructure is consuming close to 94% of the cash generated by core businesses of major AI players.
The US AI industry needs to generate approximately $600 billion in new annual revenue to justify current spending, a target it is far from reaching.
In January 2025, China's DeepSeek model's market entry reportedly caused a $1 trillion loss in US AI market value in a single day, severely impacting NVIDIA.
The US stock market's overwhelming dependence on AI
The US stock market has become disproportionately reliant on artificial intelligence, with AI accounting for 80% of all gains in the US stock market over the past three years. This concentration means that the ten largest companies in the S&P 500, now comprising over 40% of the entire index, are heavily invested in the AI sector. Consequently, owning an index fund no longer represents a diversified investment in the broad market but rather a concentrated bet on AI infrastructure. This dependence makes the market highly vulnerable to disruptions in the AI landscape, a risk that has recently been highlighted by actions from China.
China's 'distillation' strategy to undermine US AI
China is employing a strategy to circumvent US export controls on advanced AI chips by leveraging a technique known as 'distillation.' This process involves training smaller, more cost-effective AI models by using the outputs of larger, more sophisticated AI models like ChatGPT or Claude. Thousands of fake accounts are used to query these frontier models, and the captured answers are then used to teach cheaper models to replicate their behavior. While distillation is a standard technique, the issue arises when it's applied to competitor models without authorization. Anthropic has accused Chinese labs DeepSeek, Moonshot, and Miniax of running over 16 million such queries, and the White House has broadly accused China of large-scale AI model theft. Alibaba has also been accused by Anthropic of an extensive effort to clone Claude's intelligence. This strategy allows China to produce capable AI models at a fraction of the cost of US-based models, posing a significant threat to the revenue streams of American AI companies.
The AI revenue shortfall and escalating debt
Beyond external competition, the US AI industry faces a severe internal challenge: revenue is not growing fast enough to justify the massive investments made in infrastructure. An MIT study revealed that 95% of corporate AI projects failed to yield any measurable profit. This is particularly problematic given the colossal expenses involved in AI development, often described as the most expensive infrastructure buildout in history, largely financed through debt. Companies like Oracle are spending 57% of their revenue on AI, and Microsoft around 45%, with overall AI infrastructure costs consuming approximately 94% of the cash generated by their core businesses. Sequoia estimates the AI industry needs to generate an additional $600 billion in annual revenue to validate current spending, a goal that seems increasingly unattainable. This debt-fueled expansion, combined with the pressure of accelerated product cycles and market euphoria, creates a precarious situation where the industry's ability to repay its obligations is uncertain. This situation is exacerbated by rising interest rates, making debt even more expensive and shortening the runway for AI companies to achieve profitability before their debt obligations become unmanageable.
Market instability and US AI value erosion
The confluence of escalating debt, unsustainable spending, and external competition is manifesting as significant market instability. Events like the January 2025 launch of China's DeepSeek model reportedly wiped out $1 trillion in US AI market value within a single day, causing NVIDIA to experience its largest single-day loss ever. This underscores the fragility of investor confidence, which is critical for AI companies funded by debt. Broad market sell-offs, such as the one on June 23rd that erased nearly $700 billion, and Oracle's worst week since the dot-com crash, signal investor concern about how AI infrastructure is being financed. As investors begin to differentiate between companies with strong technology and revenue growth versus those reliant on debt, the market is becoming more volatile. This trend is reinforced by rising interest rates globally, which reduce market liquidity and make carrying AI debt more expensive, particularly for companies with long payoff horizons like OpenAI and Anthropic.
Calls for government intervention and regulatory capture fears
Facing immense financial pressure, AI companies are increasingly turning to the US government for support, raising concerns about regulatory capture. OpenAI has explored the idea of federal backing for its financing and has also sought tax credits and loans from the White House. While proponents argue that AI is strategically important and warrants government support, critics fear this could lead to protected industries, limited competition, and artificially inflated prices. Companies like Anthropic are advocating for binding government regulation, mandatory safety testing, and powers to block AI models deemed dangerous. However, critics argue that such regulations, while framed as safety measures, could conveniently raise barriers to entry, disproportionately harming smaller players and open-source projects, particularly those from China. This push for regulation could stifle innovation and competition, benefiting established players at the expense of the broader market and consumers.
Public skepticism and societal impact of AI buildout
The rapid expansion of AI infrastructure is also generating significant public backlash and economic strain. A large segment of the public holds a negative view of AI, with a majority expressing concern about AI data centers being built near them. Rising power bills and increased prices for everyday electronics, described by Tim Cook as a "100-year flood" and by Elon Musk as the biggest price jump he's ever seen, are directly attributed to the AI buildout. This creates a societal paradox: a technology that is already systemically important for national security and the economy is simultaneously disliked by a significant portion of the populace due to its direct costs and perceived risks. This widespread public opposition and economic impact further complicate the AI industry's sustainability narrative.
The '2008 playbook' for distributing AI debt risk
As AI companies accrue massive debt, banks are employing a strategy reminiscent of the 2008 financial crisis to offload risk. This involves packaging and selling off the risky AI debt to private credit funds, insurance companies, and pension funds. Hyperscalers alone took on over $100 billion in new AI data center debt in 2025, a significant increase from previous years. Lenders are pooling these loans to spread the risk across the financial system, including major life insurers and pension funds that already hold substantial amounts in private debt. The intention is to move risk away from banks and onto the broader economy, making it more likely that the government would intervene with financial support if the AI sector stumbles. This strategy, if successful, could create a systemic crisis where a failure in the AI market triggers a broader economic downturn, potentially necessitating a taxpayer-funded bailout.
Navigating AI investment amidst hidden risks
Given the complex landscape of AI investment, understanding where risk actually lies is paramount. AI risk often hides in seemingly safe investments such as pension funds, insurance policies, bond funds, and even diversified index funds that have become heavily weighted towards AI. History demonstrates that a strong investment thesis can be undermined by poor timing or hidden debt, especially when external factors like China's pro-consumer strategy drive down costs and increase market fragility. Investors are advised to be humble, diversify their portfolios significantly beyond AI, and consider the long game. Believing in AI's transformative potential is valid, but getting the timing wrong, especially when betting on debt, can lead to significant losses. The current situation, characterized by hidden risks, massive debt, and external pressures, closely mirrors the conditions that led to the 2008 housing collapse, suggesting a need for extreme caution and defensive investing strategies.
Mentioned in This Episode
●Software & Apps
●Companies
●Organizations
●Concepts
●People Referenced
Navigating the AI Investment Landscape
Practical takeaways from this episode
Do This
Avoid This
AI Spending vs. Revenue in Major Tech Companies
Data extracted from this episode
| Company | AI Spending (% of Revenue) | Impact/Profitability |
|---|---|---|
| Oracle | 57% | No measurable impact on profits for 95% of corporate AI projects cited by MIT study. |
| Microsoft | 45% | No measurable impact on profits for 95% of corporate AI projects cited by MIT study. |
| Across biggest players | ~94% of cash flow | No measurable impact on profits for 95% of corporate AI projects cited by MIT study. |
Cost Comparison: US Frontier vs. Chinese Open-Source AI Models
Data extracted from this episode
| Model Type | Relative Cost | Performance Benchmark |
|---|---|---|
| Top US Frontier Models | 1x (Baseline) | High |
| Chinese Open-Source Models | ~1/5 (5x cheaper) | Within a few points of top US models |
Common Questions
China is using AI distillation to create cheaper, efficient open-source models, potentially leveraging US frontier models illegally. This undercuts US AI revenue streams and exploits the US industry's reliance on debt-funded infrastructure.
Topics
Mentioned in this video
Mentioned as an example of a company that cut its AI bill by moving to cheaper open-source models, demonstrating the cost pressure on US AI services.
Accused three Chinese labs of illegally using its models for distillation and sent a letter to the Senate Banking Committee accusing Alibaba of a similar effort. Also mentioned as a company whose AI costs grew significantly.
A Chinese AI lab accused by Anthropic of illegally using US frontier models for distillation. Its model's release in January 2025 impacted US AI market value.
A Chinese AI lab accused by Anthropic of illegally using US frontier models for distillation. It developed an open-source model that Coinbase used.
Accused by Anthropic of conducting one of the largest efforts to clone AI intelligence through distillation.
Mentioned as an example of a company that exhausted its AI coding tool budget early in the year due to high costs, capping engineer spending.
Sent an internal memo warning of increasing AI spending and is mentioned as a company whose AI debt is being fueled by pension funds.
Scrapped an internal AI usage leaderboard due to employees gaming the system and driving up costs without proportional product improvement.
Experienced a significant drop in its post-IPO gains, falling 32% within two weeks, cited as a sign of market unease with AI financing.
Had its worst week since the dotcom crash due to AI buildout financing concerns and is spending a high percentage of its revenue on AI.
Agreed to invest up to $100 billion in OpenAI and also owns a piece of CoreWeave, creating circular financing concerns. Experienced the largest single-day loss by a company after China's DeepSeek model hit the market.
Received a large investment from NVIDIA, is deeply in the red, asked for government financing support (tax credits, loans), and its CEO has made statements about government being the 'insurer of last resort'.
A cloud company in which NVIDIA has invested and has committed to buying its unused capacity, contributing to circular financing concerns.
Spending a significant portion of its revenue on AI buildout and reportedly stepped back from its commitment to supply all of OpenAI's computing power.
Sponsor of the video, presented as a reliable provider for small businesses needing constant connectivity.
Sponsor of the video, offering 100% grass-fed beef sticks as a convenient, nutritious option to maintain standards.
Mentioned as a leading frontier AI model used in the process of 'distillation' for training smaller, cheaper open-source models.
Mentioned as a leading frontier AI model potentially being 'cloned' by Chinese labs through distillation, and as a model Anthropic is developing.
A company that switched from Anthropic's Claude to a Chinese open-source AI model due to escalating costs.
Conducted a survey finding that only 26% of companies have a clear view of their AI spending.
An investment firm that compared the AI industry's financing arrangements to the circular financing of dotcom bubble era companies.
Reported that major life insurers have close to $1 trillion tied up in private debt and is signaling future interest rate hikes, increasing debt costs.
Has implemented export controls on AI chips and is investigating Chinese model makers. AI companies are seeking financial support and tax credits from it.
OpenAI CEO who stated the company does not want government guarantees and taxpayers shouldn't rescue companies making bad bets, but also mused about government as the 'insurer of last resort'.
CEO of Anthropic, who advocates for binding government regulation of AI, mandatory safety testing, and the power to block dangerous AI models.
Described the rapid increase in everyday electronics prices, partly due to AI buildout, as a '100-year flood'.
Stated that the price jump in everyday electronics due to AI buildout is the biggest he has ever seen.
Facing significant risk from Chinese competition, heavy debt burdens, and lack of rapid revenue growth, potentially impacting the broader US economy.
A technique where smaller AI models are trained on the outputs of larger, frontier models to create cheaper, efficient alternatives, but potentially used illegally by China to undermine US AI companies.
Cheaper alternatives to proprietary US AI models, developed significantly by China, are attracting US customers and posing a threat to US AI revenue.
The buildout and operational costs associated with AI data centers are a major driver of the industry's financial strain, with 71% of Americans opposing them being built nearby.
The risk that AI companies, by seeking government support and regulation, might unduly influence policy to benefit themselves and stifle competition.
More from Tom Bilyeu
View all 113 summaries
123 minThe Bleak Jobs Report, Divisive Messages On July 4th & A US World Cup Scandal
130 minThe DSA Is More Sinister Than You Think, Caleb Hammer Explains Taxes, & Alex Karp Pops The AI Bubble
30 minThere's No Plan To Pay The $40 Trillion Debt — The Plan Is To Steal It From Your Savings
122 minIran Is About to Control the World's Oil — and You'll Pay for It
Ask anything from this episode.
Save it, chat with it, and connect it to Claude or ChatGPT. Get cited answers from the actual content — and build your own knowledge base of every podcast and video you care about.
Get Started Free