Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsInvestor Roger McNamee’s warning is about a mismatch: technology companies are committing large sums to AI models and infrastructure, while revenue from software sold to end users may not grow enough to support those investments. He argues that the gap could force some companies to write off investments—but that is a forecast, not evidence that an industry-wide collapse has happened or is certain.
What McNamee’s warning means
McNamee made the argument in his Guardian opinion essay, “AI investors are in for a rude awakening,” published September 24, 2025, and modified September 25. A September 26, 2025 Futurism story summarized the warning behind the headline about the AI industry collapsing for a basic financial reason.
The concern is not simply that AI is expensive. Companies are spending on large language models (LLMs), data centers, chips, and cloud infrastructure on the expectation that customers will eventually pay enough for AI products and services to justify the cost. McNamee argues that end-user license revenue is not keeping pace with that investment.
The financial gap at the center of the thesis
Investment is not the same as customer revenue
McNamee estimates that the industry will have invested about $717 billion over three years by the end of 2025 in LLM AI and supporting infrastructure. This is his estimate in a 2025 opinion essay, not an audited or independently confirmed final total.
#1 Best Overall
His comparison is between that scale of investment and the much smaller revenue he says is coming from licenses for end-user AI software. The distinction matters: capital spending can happen now, while the customer revenue expected to repay it remains a future projection. If that revenue falls short, companies may struggle to earn an adequate return on their investment.
Why activity across the industry may not prove demand is sufficient
Spending on AI infrastructure can create revenue for chip suppliers and cloud operators, while investment can also flow to startups building AI products. McNamee points to these channels as part of the industry’s financial activity. His interpretation is that money moving among companies does not, by itself, establish that final customers are paying enough for AI software to support the overall level of investment.
That is not the same as saying that all AI revenue is circular or fictitious. The narrower point is that supplier sales, cloud revenue, and startup funding are different from evidence that end users are buying software at a scale sufficient to make the investment economics work.
Why he thinks competition raises the risk
McNamee says several major technology firms are pursuing leadership in LLMs, each with ambitions he characterizes as seeking a global monopoly. “Each big tech company needs a global monopoly in AI to sustain their success and market value,” he writes, adding: “They are not all going to get one.”
Rank #3
His essay names Google, Amazon, Meta, xAI, and Microsoft/OpenAI among five or six major US LLM programs. The count is approximate and depends on how programs and partnerships are counted. He also mentions Nvidia, Apple, Anthropic, and other companies in the broader discussion.
The financial risk follows from the combination of large investments and a contest for dominance: if only a small number of companies can become highly profitable leaders, some of the other players may not earn enough to justify what they spent. McNamee therefore predicts that several could eventually need to write off investments. He presents this as a possible outcome, not a documented collapse.
Rank #4
How to separate the warning from what it establishes
| Financial question | What McNamee argues | What the essay does not establish |
|---|---|---|
| Capital spending versus end-user revenue | Investment in LLMs and supporting infrastructure greatly exceeds revenue from end-user software licenses. | An independently audited industry-wide comparison of spending and license revenue. |
| Forecasts versus reported revenue | Future customer revenue is expected to support current investment. | That projected revenue will arrive, or that it has already done so. |
| Supplier income versus final-user payments | Chip sales, cloud services, and startup investment contribute to the industry’s financial flows. | That activity among suppliers and companies proves sufficient final-customer demand. |
| Possible winners versus capital at risk | Several major firms are competing for leadership, but not all can achieve the dominance McNamee says they seek. | Which firms will succeed, which will lose money, or whether any particular investment will be written off. |
McNamee, identified in the Guardian byline as managing director at Elevation Partners and an early-stage investor in Google and Facebook, summarizes his view this way: “The huge gap between capital investment in infrastructure and end user license revenue from AI software is not sustainable.” That is his assessment in an opinion essay, rather than a neutral finding established by the newspaper.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What would make the warning come true
The thesis depends on several things happening together: spending remains high, end-user revenue fails to catch up, and competition prevents enough firms from earning returns that justify their investments. If those conditions hold, McNamee’s predicted write-offs become more plausible. If customer revenue grows enough to support the investment, the gap he describes may narrow instead.
The essay does not independently verify the $717 billion estimate against company filings or an underlying dataset, nor does it establish whether the predicted write-offs occurred. Determining how the forecast has played out requires checking subsequent company filings and data on revenue and capital expenditure.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
