Free tools Windows power users keep installed
One-click scans. No signup required.
Michael Burry’s bearish case is not simply that AI is useless. It is that companies may be spending heavily on servers, GPUs and networking equipment that lose economic value faster than their accounting schedules assume, leaving AI infrastructure returns weaker than reported earnings suggest. That argument could be wrong: equipment can remain productive for years, cash generation matters alongside accounting profit, and today’s investment could support durable recurring revenue. None of those points, on its own, proves that current spending or stock valuations will pay off.
What does Burry’s bearish case actually say?
Burry argues that the economics behind the AI buildout may be less attractive than the market expects. Hyperscalers are committing large sums to data centers and computing equipment. If that equipment becomes economically obsolete sooner than companies’ accounting assumptions allow, its cost should be recognized over fewer years. Faster depreciation would reduce reported earnings and could make returns on AI infrastructure look weaker.
Depreciation is an accounting method for allocating a tangible asset’s cost across its estimated useful life. A longer useful-life estimate spreads the expense over more years and generally raises near-term reported earnings compared with a shorter estimate. The key question is not merely how long equipment remains on a balance sheet, but how long it remains productive and economically valuable.
In 2025, The Motley Fool reported Burry’s estimate that depreciation across the industry could be understated by about $176 billion over 2026–2028. The same report attributed to him estimates that Oracle earnings could be overstated by about 27% and Meta earnings by about 21%. These are Burry’s projections, not confirmed accounting errors or losses that have already occurred.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →#1 Best Overall
What do Scion’s reported put positions show?
Scion Asset Management’s SEC Form 13F reported put positions as of September 30, 2025, tied to one million Nvidia shares and five million Palantir shares. The filing table listed values of $186.58 million and $912.1 million, respectively. Those are reported values associated with the underlying securities, not the option premiums Scion paid, a measure of maximum profit, or evidence of a realized return.
A 13F is a delayed disclosure with limited scope. It does not establish what happened to the contracts after the reporting date, whether Scion held other offsetting exposure, or what Burry’s position is now. The SEC filing also states that the commission has not determined whether its information is accurate and complete. It is therefore accurate to say the filing reported puts as of September 30, 2025—not that Burry is currently short Nvidia or Palantir.
Why might Burry’s depreciation argument be wrong?
Useful life on paper is not the same as productive life
Newer hardware can be faster without making older hardware worthless. Nvidia management said on an earnings call, as quoted in The Motley Fool’s November 2025 analysis, that A100 GPUs “we shipped six years ago are still running at full utilization today.” That company statement supports the possibility that some older GPUs remain heavily used; it does not establish the useful life, efficiency, or resale value of every GPU or data-center system.
Rank #2
- Ideal for Gifting
- Ideal for a bookworm
- Compact for travelling
The distinction cuts both ways. Hardware can remain in use while becoming less efficient or less valuable than expected. But the arrival of a newer generation does not by itself prove that existing equipment is obsolete or that its accounting life is too long.
Recommended Free Tools
Accounting earnings are not the whole financial picture
A longer depreciation schedule raises reported earnings in the near term, but that fact alone does not demonstrate that a company is in financial distress or that its investment is failing. The counteranalysis in The Motley Fool’s November 2025 report points to cash generation, tax and cash-flow consequences, and the use of measures such as EBITDA alongside net income. Large technology companies may be able to finance substantial investment from their cash generation. That weakens a simple leap from “earnings are higher under a longer depreciation schedule” to “the company cannot afford its investment.” It does not show that the investment earns an adequate return.
Infrastructure could generate revenue for years
Hyperscalers may be investing now to sell cloud capacity, AI services and recurring customer products in the future. If that spending creates durable demand and revenue, a near-term depreciation charge does not settle whether the investment will pay off over its full life. The reverse is also true: anticipated demand is not proof that each data center, model or hardware purchase will be profitable.
The disagreement is about the return on investment, not only whether AI has useful applications. Palantir’s third-quarter 2025 results offer one dated example of growth alongside bearish scrutiny: The Washington Post reported revenue of $1.18 billion, up 63% year over year, and a 33% profit margin. Those figures describe one quarter at one company; they do not establish that its valuation—or the broader AI buildout—is justified.
Companies have changed useful-life estimates in different directions
The Motley Fool’s November 2025 analysis, drawing on SEC filings, reported the following changes. They show that useful-life assumptions have not moved uniformly across large technology companies.
| Company | Reported change | What the source says about the reason |
|---|---|---|
| Alphabet | Extended server useful lives from four to six years in 2023; extended certain network equipment from five to six years. | The cited summary reports the changes; it does not state a specific reason here. |
| Amazon | Shortened server useful lives from six years to five in 2025. | The cited analysis attributed the change to faster technology development, particularly in AI and machine learning. |
| Meta | Moved certain server and network assets to 5.5 years in 2025, from four to five years. | The cited summary reports the change; it does not state a specific reason here. |
| Microsoft | Extended server and network equipment useful lives from four to six years in 2022. | The cited summary reports the change; it does not state a specific reason here. |
| Oracle | Extended server useful lives from five to six years in 2025. | The cited summary reports the change; it does not state a specific reason here. |
These filing-based changes do not prove that any company’s estimate is right or wrong. They do show why it is misleading to describe every hyperscaler as following the same depreciation policy—or to treat a change in policy as a direct measure of actual hardware performance.
Rank #4
Can AI be valuable while AI stocks are overpriced?
Yes. Evidence that customers are adopting AI, that a company is growing, or that older equipment remains in use can support the view that the businesses are durable. It does not show that investors’ expectations are reasonable or that the money going into infrastructure will earn a sufficient return.
A useful way to keep the debate clear is to separate four questions:
- Asset life: How long is equipment actually useful and economically valuable, compared with its accounting life?
- Financial performance: What do cash generation, taxes, depreciation and reported profit show when considered together?
- Investment payoff: Will future cloud, AI-service and product revenue justify current capital spending and operating costs?
- Stock price: Do expected returns justify what investors are paying, even if the technology and businesses are successful?
The available figures here do not provide a synchronized comparison of company valuations sufficient to say whether the stocks are cheap or expensive. The technology can work, businesses can grow, and a stock can still disappoint if its price already assumes too much success.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallBest Value
- It can be a gift option
- Comes with secure packaging
- Helpful in various ways
What is Burry saying about the AI boom beyond depreciation?
A September 2026 Motley Fool report described Burry’s broader concerns about the AI industry and calls to slow its development. It attributed to him the view that large language models are not equivalent to AI or artificial general intelligence, that competition could erode incumbents’ advantages, that IPO promotion may lean on hype, and that calls to slow development could conceal slowing growth or delayed IPOs. These are Burry’s interpretations of industry dynamics and company motives, not independently established explanations of why particular people or firms have taken public positions on AI safety.
The same report characterized planned 2026 AI-infrastructure spending by the “Magnificent Seven” at about $750 billion. That is a media report’s description of planned spending, not a verified total of actual spending in the evidence available here. To evaluate Burry’s broader argument, compare what companies do and report—capital commitments, customer demand, recurring revenue, profits or losses, model competition and IPO plans—with their public claims. Those indicators can illuminate business conditions; they cannot establish private intentions on their own.
What would show whether the bearish case is holding up?
The most useful evidence will be results over time rather than a single disclosure or estimate. Watch whether actual equipment use and replacement patterns support the useful lives companies report; whether AI-related revenue and recurring customer spending grow enough to cover infrastructure costs; and whether cash generation and returns keep pace with capital commitments. Also distinguish changes in depreciation estimates from changes in the underlying economics: either can matter, but they answer different questions.
Burry’s public reputation partly comes from anticipating problems in the mortgage-backed securities market, a story depicted in Michael Lewis’s The Big Short. That history explains why investors pay attention to his warnings; it does not prove that his current AI thesis is correct. His 2025 put disclosures likewise show a dated reported position, not whether the thesis later paid off.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →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.
