The Tech Stocks Going To ZERO: Bankruptcies ‘Only A Question Of Time’| George Noble

Summary

George Noble argues that debt-funded AI infrastructure is vulnerable to a financing downturn, with highly leveraged “neocloud” companies—especially CoreWeave—at risk of insolvency. He says investors should focus on balance sheets, cash flow, and valuations rather than AI narratives or falling share prices. He also warns about risks in life insurers and sees opportunities in energy, precious metals, and shipping.

These are Noble’s opinions and forecasts, not established outcomes.

Shorts and AI-sector risks

• CoreWeave is Noble’s largest short. He says he initiated the position around $107, added to it around $81, and recently doubled it. He believes bankruptcy is a meaningful possibility.
  • The discussion cites about $55 billion of total debt; the host also refers to roughly $35 billion of debt, leaving the exact breakdown unclear.
  • Noble cites an estimated 8.9% return on invested capital (ROIC) over the next 1.5–3 years, versus borrowing costs he says are higher. Some bonds were said to be yielding 15%.
  • He questions the economics of buying and renting out rapidly depreciating GPUs: debt and depreciation persist, while long-term demand, chip useful life, and sustainable rental economics remain uncertain.
  • The host says OpenAI holds up to $22.4 billion of CoreWeave contracts and cites HSBC estimates. Noble says CoreWeave’s customer and financing concentration add to its risk.
• Noble sees CoreWeave as a sign of a broader AI infrastructure financing problem, not an isolated case. He expects other neocloud operators to suffer if it fails and says the bond market may be signaling more risk than equity investors are pricing.
• He argues that the AI investment boom depends heavily on external funding for OpenAI and Anthropic, which he describes as cash-burning businesses. If investors become less willing to fund successive rounds at higher valuations, growth and valuations could fall.
  • He says OpenAI’s IPO has been postponed to 2027; Anthropic’s timing remains uncertain. The host cites prediction markets showing a greater-than-60% chance of an Anthropic IPO announcement by year-end.
  • Noble says Anthropic’s reported valuation ambitions had shifted from about $2 trillion to $1.5 trillion. He argues that 45% revenue growth would not support such a valuation and says growth closer to 150% would be needed.
• Noble questions AI’s addressable market and pricing. He says an individual subscription priced around $20 per month may be uneconomic if the true cost is closer to $200 per month. He also argues that corporate adoption may be limited by data-security and compliance concerns, and that cheaper models—he claims Chinese competitors can be 95% cheaper—could intensify price pressure.
• He warns that heavy capital expenditure could leave hyperscalers with little or no profit by 2030, as depreciation on data-center investment rises. He also cites a report that Amazon had about $200 billion of chips sitting in warehouses. In the interview, this was presented as a reported example of possible over-ordering and deployment bottlenecks.
• Noble compares the AI boom to the 2006 housing-market warning signs before the Global Financial Crisis. His concern is not necessarily that the technology is unreal, but that financing depends on continued asset appreciation and fundraising. He also compares the risks to the dot-com era, saying that a technology can succeed while investors in particular companies still lose heavily.
• He argues that related-party financing and accounting can make reported revenue and profits a poor guide to underlying economics. He describes the current boom as potentially worse than the dot-com bubble because of its scale and financing structures, though this is his assessment.

Other short positions and valuation views

• Tesla: Noble says he initiated his short around $475–$485 and that it was approximately 25% ahead, while the broader market was up roughly 10–15%. He says Tesla’s cash-flow burn is intensifying and believes its fundamental value is around $30 per share, while cautioning that the stock need not actually reach that price.
  • He dismisses the Roadster presentation and robotics narrative as insufficient to support the valuation.
  • He estimates a hypothetical robotics business might be worth $30–$40 billion, too little to justify a market capitalization of roughly $1.4–$1.5 trillion.
• SpaceX: Noble also describes it as a short and says it has a cash-flow problem. He reports adding to the position after a price decline and subsequent rebound.
• Oracle: Noble warns about its borrowing and AI-related capital spending. The host cites negative free cash flow for the fiscal year ending in May 2026, around $260 billion in data-center leases, and a possible move toward junk-grade credit. Oracle’s share price is cited as around $140. Noble says a lower share price alone does not make a stock attractive if fundamentals are deteriorating.
• Noble says semiconductor stocks are already weakening and could face further pressure if neocloud spending and chip demand fall. The host cites CoreWeave as buying roughly 8% of Nvidia’s chips—a figure the host says is not precisely disclosed—and notes Nvidia’s backing of CoreWeave.
• Noble says a failure of CoreWeave would not necessarily be a banking-system crisis. He sees more exposure in private equity, private credit, endowments, and pension funds. He nevertheless warns that lenders or investors left holding Oracle debt could face losses.

Life-insurance concerns

• Noble alleges that some life insurers and annuity providers have taken excessive exposure to illiquid securities, private credit, private equity, and related-party transactions, and that oversight is inadequate. He argues that insurers’ long-dated liabilities can obscure problems for years.
• Companies he names include Athene, Prudential, Equitable, and MetLife. These are allegations and forecasts made by Noble in the interview.
• He says policyholders at insurers he considers bad actors should move their money to a stronger insurer, warning that policyholders may rank behind creditors in a failure.
• He recommends that viewers hear Tom Gober, whom he describes as a longtime insurance investigator.

Market and macroeconomic outlook

• Noble agrees with the interview’s interpretation of a statement by Kevin Warsh: monetary policy has tightened synchronously in many countries, and global inflation remains a shared concern. Noble expects the cost of capital to remain a key risk.
• He says markets, rather than the Fed, ultimately determine financing conditions. He cites a 10-year Treasury yield near 4.70% and criticizes a bond-buyback increase announced by Treasury Secretary Scott Bessent as too small to materially change market conditions. The transcript gives inconsistent figures for the buyback increase.
• Noble says technology represents about 40% of the S&P 500, with some broader estimates suggesting 50–60% of the market trades in sympathy with tech. He says many individual stocks are already down roughly 15% from their highs, with a substantial number down 20–30%, even while major indexes appear relatively resilient.
• He expects volatility to increase after the election, arguing that incentives to support markets may diminish afterward. He also claims the U.S. government is trying to hold down oil prices and support stocks; this is his view.
• Noble says the AI and data-center buildout could slow sharply if financing costs rise, permitting delays persist, or political opposition to data centers grows.

Long positions and portfolio approach

Noble favors companies with less debt, lower valuations, near-term cash flow, and exposure to inflation. He says his long ideas include:

• Energy: Schlumberger, Valaris, Harbor Energy, and other energy companies. One Canadian energy holding is not intelligible in the transcript.
• Precious metals and miners: Gold-mining stocks; SSRM, which he calls attractive at about 7–8 times earnings; a silver play identified as “MAKO” in the transcript; and another mining name identified as “AUGO.” The latter two ticker or name renderings may be transcription errors.
• Shipping: Star Bulk Carriers (SBLK) and other shipping stocks.

He reports a 67% raw long–short spread, with an annualized figure he estimates at about 120%. He cautions that performance is unlikely to continue at that rate and says the results involved some luck.

His stated framework:

• Evaluate fundamentals, valuation, cash flow, and the financing structure, not just the share price or prevailing narrative.
• When a position moves, ask whether the investment thesis is improving or deteriorating. Do not automatically cover a short because the stock has fallen or sell a long simply because it has risen.
• In a higher-cost-of-capital environment, favor low-debt companies, reasonable valuations, short-duration cash flows, and businesses that may benefit from inflation.
• Judge investment performance by the process, not only the scoreboard.

Assets, companies, and instruments mentioned

• Equities and companies: CoreWeave; Tesla; SpaceX; Oracle; Nvidia; Microsoft; Amazon; OpenAI; Anthropic; Schlumberger; Valaris; Harbor Energy; Star Bulk Carriers; Athene; Prudential; Equitable; MetLife.
• Tickers explicitly or apparently spoken: NBIS, CIFR, SSRM, SBLK, plus “ORC” (likely a transcript rendering of Oracle’s ticker, ORCL) and the uncertain “MAKO” and “AUGO.”
• Other assets and instruments: AI chips and GPUs, data centers, corporate bonds, credit default swaps (CDS), private credit, private equity, U.S. Treasuries, oil, gold, silver, the S&P 500, and a referenced Australian IPO.
• Sectors: AI infrastructure, semiconductors, technology, life insurance, energy, precious-metals mining, and shipping.

Disclosures and cautions

• The sponsor segment describes GigaStar and its channel revenue tokens, and says early-stage investment carries risk; viewers are told to read the offering materials.
• Noble cautions that his reported performance is unlikely to be repeated. The interview includes strong investment views and forecasts; it does not establish that the predicted bankruptcies or market declines will occur.

Presenters and sources

Presenters: David Lin (host) and George Noble (guest), Managing Partner of Noble Capital Advisors.

Sources or commentators referenced in the discussion: Kevin Warsh; Scott Bessent; Tom Gober; Peter Lynch; Peter Barron; Stan Weinstein; Don Cox; Julian Garren (name as rendered in the subtitles); and a Canadian writer identified only as “Jay.” Noble also refers to George Soros and Stanley Druckenmiller in discussing the 1992 UK currency crisis.