This article is crafted based on publicly accessible information and serves solely as a platform for informational exchange, not as investment guidance.

According to an exclusive report by NOTUS on July 6, the U.S. Treasury Department has completed an internal report on risks within the artificial intelligence market several weeks ago, and it is now pending formal approval. Authored by professional analysts, the document can be distilled into a single, stark sentence: The AI market is echoing the dot-com bubble of 2000, and its collapse could reverberate throughout the entire economic system.

(Note: NOTUS is a news agency specializing in U.S. politics and policy coverage, established in 2024 by a group of seasoned journalists renowned for delivering in-depth, exclusive insider news from Washington.)

However, as the report awaited endorsement, Treasury Secretary Scott Bessent stood at a podium in New York, lauding tech behemoths for their $750 billion investment in AI development this year. During the recent G7 summit, he emphasized to other leaders that AI's paramount risk lies not in security or employment but in China's lead over the U.S.

The policy discourse surrounding AI is notably disjointed.

01 Insights from the Report

According to the internal document obtained by NOTUS, Treasury analysts have constructed a clear causal chain in their report. AI firms are more deeply entrenched in the U.S. economic fabric than internet companies were, relying on intricate financing mechanisms and substantial infrastructure investments. Should financial conditions worsen, productivity targets be missed, or supply chain or electricity bottlenecks hinder growth, the entire system could face significant shocks.

The report's evaluation is quite detailed: An AI downturn would send "shockwaves throughout the economic ecosystem." The stock market, private credit markets, corporations financing data center construction, cloud service providers, chip manufacturers, and utility companies all form part of this transmission chain.

Analysts particularly highlight that the AI industry's concentration far surpasses that of the internet era, with a handful of firms dominating the market. These giants are highly interconnected and deeply intertwined with various markets. If investment dries up or demand wanes, the chain reaction will not be confined to the tech sector.

The most alarming aspect of the report is its deep-seated judgment: Unlike the dot-com bubble, which primarily involved retail investors, institutional investors are heavily involved in the current AI boom. When retail bubbles burst, they harm stock markets and household wealth; when institutional bubbles burst, they threaten the stability of the entire financial system. Large banks, hedge funds, and private credit institutions are the primary financiers of this AI venture. Risks are not dispersed but concentrated in a few institutions deemed "too big to fail."

Of course, the report does not claim that the bubble will inevitably burst. According to NOTUS, analysts acknowledge fundamental differences between AI firms and those during the dot-com bubble: They are more mature, some already profitable, with healthier balance sheets.

The report even offers an optimistic outlook: If productivity growth expectations materialize and commercialization pathways succeed, the so-called bubble may never burst. However, this also implies that the stability of the entire financial system now hinges on AI delivering on its promises.

02 The Narrative from the Podium

Bessent's speech in New York presented a contrasting narrative. Citing the productivity miracles of the internet era, he challenged the audience: Can we at least achieve that? Can we do more?

He did not address bubbles, risk exposures, or concentration. His focus was solely on speed.

This tone cannot be dismissed as mere political duplicity. The U.S. government's AI strategy is built on a single premise: AI is a race that cannot be lost. The cost of losing to China outweighs that of any bubble bursting.

Under this premise, all discussions about risk, leverage, and concentration are relegated to secondary issues. Much like how the Manhattan Project did not pause to debate the potential self-harm of atomic bombs, or how the space race did not halt to calculate the return on investment for moon landings. Under the logic of competition, risks are downgraded to temporary setbacks that must be endured before achieving success.

A Treasury spokesperson issued a nearly contradictory statement in response. On one hand, he told NOTUS that the report was "unreviewed" and "does not represent Treasury policy or views." On the other hand, he reiterated the Treasury's official stance: "Artificial intelligence will be a key driver of America's new golden age." The subtext was clear: We acknowledge the existence of that analysis in the drawer, but we choose not to look at it because a higher goal demands our attention.

If a historical parallel is sought for this choice, the closest is not the dot-com bubble but the U.S. housing market before the 2008 financial crisis. Around 2005, analysts within the Treasury and Federal Reserve warned about systemic risks from subprime mortgages. Those reports were equally clear, with equally complete transmission chains, and were equally locked away in drawers.

The dominant narrative then was also: Housing prices always rise, financial innovation disperses risks, this time is different. The difference is that before the subprime crisis, some officials still publicly discussed risks and pushed for regulation. Today, even that voice has nearly vanished.

03 A Towline or a Noose?

The outside world is not entirely oblivious. The Bank of England and the International Monetary Fund president have publicly expressed concerns about AI overvaluation and its risks to the economic system.

Domestically, Capitol Hill is taking action. According to NOTUS, Senator Elizabeth Warren, a senior member of the Senate Banking Committee, proposed a bill last month requiring financial firms to disclose non-public data related to AI development to the Treasury and mandating the Treasury to submit a special report on the financial system's various risk exposures to AI. Warren's wording was direct: "AI and Big Tech companies increasingly rely on opaque debt forms and balance sheet magic to support their trillion-dollar AI builds."

This bill itself is an indictment of the status quo. Markets do not know AI companies' true leverage ratios, how many cross-guarantees are hidden in data center financing structures, or how many private credit contracts would simultaneously trigger default clauses if AI computing rental prices fell. The only entity capable of forcing disclosure of this information—the Treasury—is locking the report in a drawer while saying on the podium to maintain the lead.

NVIDIA's GPUs, cloud providers' data centers, startups' valuations, and Big Tech's capital expenditures—the prices of these assets are intertwined by the same thread of expectations.

If AI companies deliver on productivity growth promises, this thread becomes a towline, pulling all assets upward. If they fail, it becomes a noose, with one asset's decline transmitting through the same thread to all others.