Why Hasn’t The AI Bubble Burst (Yet)?

Summary

The video argues that the AI boom may have bubble-like risks but has not burst because spending and financing continue to support it. Large technology companies are committing enormous sums to AI infrastructure, while chipmakers, private investors, and corporate partners help fund AI companies—even when some AI firms are deeply unprofitable. The video warns that upcoming IPOs could shift risk from early investors to public-market investors, including people with broad-market retirement funds.

The subtitles use “bananas” as a stand-in for currency; figures below are presented as dollar amounts. The video’s figures and reported claims are not independently verified here.

Why the boom is still supported

• Infrastructure spending: Alphabet, Amazon, Microsoft, and Meta are said to plan about $725 billion in AI spending this year, mainly for chips, servers, and data centers.
• Government backing: The video cites the Stargate plan, announced with Sam Altman and Donald Trump, as a proposal to invest up to $500 billion in U.S. AI infrastructure. It also describes further White House support for data-center construction.
• Strong demand for compute: Nvidia is portrayed as benefiting from demand it cannot yet meet. The video says Nvidia’s revenue for the three months from May to July was $96.2 billion, more than double the year-earlier figure, with about $89 billion from data-center chips. It characterizes Nvidia as retaining roughly 75 cents of each revenue dollar after chip-production costs—a gross-margin-style comparison, not net profit.
• Revenue and commitments at major platforms: Microsoft’s AI business is described as generating $37 billion a year, up 123%. Google is said to have signed $467.6 billion in contracts, largely for customers to rent computing capacity in future years.
• AI usage: ChatGPT is reported to have about 900 million weekly users, but only roughly 5% are said to pay. The video argues that user scale has not yet translated into profitability for AI model developers.

Profitability concerns and financing loop

• OpenAI: The video says that in 2025, OpenAI lost about $1.60 for every $1 of revenue from operations. It compares this with a reported loss of $2.37 per revenue dollar the year before, while also saying operating losses rose from about $8.8 billion to nearly $21 billion.
• Anthropic: Reportedly lost about $8 billion in 2025 on $4.6 billion of revenue.
• Circular financing concern: The video describes a loop in which Nvidia invests in OpenAI; OpenAI buys Nvidia chips and commits to a $300 billion cloud deal with Oracle; and Oracle uses funds to buy Nvidia chips. It says Nvidia announced similar deals totaling more than $540 billion that year and was discussing guarantees for $250 billion in loans for Sam Altman’s data centers.
• The concern is that apparent demand for AI infrastructure may partly depend on financing among companies that are also suppliers, customers, or investors in one another. The video compares this risk with Nortel in the early 2000s, when customer financing reportedly contributed to problems after customers failed.
• Oracle-related reports: The video cites reports that some Oracle data centers for Sam Altman were about a year behind schedule and that expansion plans for the main Texas campus were halted after funding talks stalled. Oracle reportedly disputed those claims. The video also says Oracle’s debt-default insurance cost rose to its highest level since 2009.

IPOs and the potential transfer of risk

• The video says OpenAI, Anthropic, and xAI are pursuing public-market listings at valuations that could reach the trillions, despite the losses it describes.
• It argues that IPOs would expose the companies’ financials and valuations to the broader market. Investors could benefit if the numbers justify the prices, but public investors could face losses if they do not.
• Bank of America is cited as calling the IPO wave a “massive transfer of risk” from early investors to later buyers. The video uses SpaceX, which it says raised $75 billion through an IPO, as an example of the scale of public-market fundraising.

Concentration and market-wide risk

• The video points to South Korea as a cautionary example: Samsung and SK Hynix reportedly made up about half of the country’s stock market. After AI enthusiasm drove their shares higher, retail investors—including some using borrowed money—bought in heavily. In July, SK Hynix reportedly posted a sixfold increase in profits but still missed expectations, after which the market fell and many investors suffered losses.
• It contrasts a South Korean market valued at about $4 trillion with a U.S. market worth about $75 trillion. The seven largest U.S. technology companies are described as worth nearly $24 trillion—about six times the entire South Korean market.
• The video says around 58% of U.S. households own stocks, often through 401(k) or Roth IRA retirement accounts. It estimates that an investor holding an S&P 500 fund has roughly one-third of the investment in the same seven companies.
• It acknowledges that the U.S. market is deeper and that the major companies generate substantial revenue, but argues that their scale means a sharp fall could have global consequences.

Key takeaways and cautions

• The boom is being sustained by large corporate spending, government support, demand for computing capacity, and private financing—not necessarily by the profitability of every AI company.
• The video’s central risk scenario is that a slowdown, disappointing results, or tighter funding could expose the gap between AI infrastructure spending and the earnings of some AI developers.
• The presenter says AI itself is likely to persist, but compares the risk to the dot-com crash, in which the underlying technology survived while many companies did not.
• The presenter says they are not an economist and do not have access to OpenAI’s financial statements. The subtitles include no explicit “not financial advice” disclaimer.

Assets, companies, sectors, and instruments mentioned

• Companies: Nvidia, Alphabet/Google, Amazon, Microsoft, Meta, OpenAI, Anthropic, Oracle, xAI, SpaceX, Samsung, SK Hynix, and Nortel.
• Markets and instruments: S&P 500, U.S. and South Korean equities, IPOs, 401(k) plans, Roth IRAs, and debt-default insurance.
• Sectors and themes: AI, semiconductors, cloud computing, data centers, robotics, and technology stocks.
• Tickers: No ticker symbols are explicitly stated in the subtitles.

Presenters and sources mentioned

• Presenter/narrator: “Monkey,” from Financial Delusion (the channel is identified in the video context as MonkeyExplains).
• Named commentators: “Steve” and “Michael,” whose identities are not specified in the subtitles.
• Other named figures: Jensen Huang, Elon Musk, Sam Altman, Donald Trump, and “Marcus” (surname not specified).
• Sources or claims referenced: Bank of America; reported company financials and contracts; news reports concerning Oracle data-center schedules and Stargate expansion; and the South Korean market example. The subtitles do not provide specific report titles or links.