In recent weeks, the spotlight in markets has turned to two striking developments: the public scepticism of Michael Burry (of The Big Short fame) toward the Artificial Intelligence (AI) rally, and the reported reduction of a major stake in Nvidia Corporation (NASDAQ: NVDA) by SoftBank Group Corp.
Together, these moves serve as a useful prism through which we can assess whether the hype around AI is cooling, or at least warranting a more cautious stance from investors and traders.
Michael Burry – who really became known for his early warnings around subprime mortgages during the Global Financial Crisis (GFC) – has flagged concerns about AI-driven valuations and corporate disclosures.
Meanwhile SoftBank’s exit (or trimming) of its Nvidia position has sparked questions about institutional timing, profit-taking and the broader sentiment underlying the AI trade.
In this article, we’ll examine what these two high-profile events could signal for market participants, including those monitoring AI-related market trends in recent years.
Key Points
- Michael Burry’s caution on AI valuations highlights growing concern about stretched assumptions and concentrated positioning in major tech names.
- SoftBank’s exit from its Nvidia stake signals a potential shift in institutional sentiment as capital moves from mature AI winners to newer opportunities.
- Together, these developments suggest that the AI trade may be entering a phase where market enthusiasm gives way to closer scrutiny of fundamentals and expectations.
The Context: AI Stocks and Market Euphoria
The last two years have seen a remarkable run for companies tied to generative AI, high-performance computing and the underlying chip infrastructure that enables large-language models (known as LLMs) and advanced machine-learning applications.
Firms such as Nvidia, Advanced Micro Devices (AMD) (NASDAQ: AMD) and Super Micro Computer Inc (NASDAQ: SMCI) have surged, driven by investor optimism that we are entering a “new paradigm” of computing.
Nvidia’s GPUs remain the architecture of choice for many hyperscale data centres; AMD has leveraged its data-centre business to ride that wave; and Super Micro has benefited from increased demand for turnkey systems.
The overarching market narrative of generative AI applications (chatbots, image generation, enterprise AI), marrying strong chip demand, has unleashed a broad-based rally in tech. Investment funds, both large and small, have piled in. Retail investors have followed suit, often through thematic ETFs or “AI basket” plays.
Valuations have soared. Multiples once reserved for high-growth software companies migrated into hardware, semiconductors and infrastructure segments. In many cases the belief was that today’s huge capex numbers will underpin revenue growth for years to come, and pricing models were built on that assumption.
That level of exuberance, when viewed in hindsight of previous technology cycles, raises questions about sustainability. Global funds and retail accounts got heavily concentrated in AI-linked trades throughout 2024–2025.
Of course, any degree of concentration like this increases vulnerability to shifts in sentiment, regulatory risk or execution disappointments. When too many participants are on one side, the potential for a sharp unwind grows.
Michael Burry’s Scepticism on AI Valuations
So, what about Michael Burry? He has earned a reputation as a contrarian investor. His early-2000s bets on sub-prime mortgages in the US and his famous “Big Short” recognition placed him in the narrative of identifying systemic risk early.
His style has been famously candid and straight to the point; dig into fundamentals, challenge consensus, and look for structural imbalances.
Recently, Burry has turned his attention to the AI space. He reportedly amassed large put positions against Nvidia and Palantir Technologies. In one report, his fund bought around US$187 million of puts on Nvidia and about US$912 million of puts on Palantir [1].
Separately, he has publicly criticised large tech companies, including Oracle Corp (NYSE: ORCL) and Meta Platforms Inc (NASDAQ: META), for extending depreciation schedules on AI hardware beyond what he sees as realistic. He has argued that this move amounts to one of the more common “frauds” of the modern era.
Burry’s reasoning is multi-fold and touches on several key points:
- He views current valuations as stretched. When firms are trading at very high multiples based on overly optimistic assumptions, the margin of safety shrinks.
- He sees herd behaviour wherever he looks. When many investors assume the same “new paradigm” story, risk is ignored and investors pile in.
- He draws parallels to past tech bubbles, where infrastructure investment and hype outpaced revenue fundamentals.
- He highlights accounting and execution risk. For example, if AI hardware is being depreciated over five or six years when the technology lifecycle is much shorter (he argues two to three years for upgrades). That mismatch may inflate earnings and make the business appear more profitable than the underlying economics justify.
For traders and portfolio managers this is a sharp wake-up call: one of the most notable value-seeking investors is pointing to the AI trade as having multiple “bubble elements”.
Whether one agrees with his timing or not, it challenges the notion that the trade is purely risk-free momentum right now. If someone like Burry is betting against you, it’s worth asking why.
SoftBank’s Nvidia Stake Sale: A Strategic or Sentiment Shift?
Meanwhile, SoftBank’s involvement with Nvidia has been significant historically. Through its much-publicised Vision Fund, SoftBank backed many tech ventures, including those centred on AI. Nvidia has clearly been one of the key beneficiaries of the AI infrastructure wave that SoftBank has long championed.
However, recent filings and commentary show that SoftBank sold its entire roughly US$5.8 billion stake in Nvidia [2]. The sale was disclosed as part of a strategy to raise capital for further investments in AI ventures such as OpenAI.
SoftBank stressed that the move was not an abandonment of AI but rather a redeployment into other AI-oriented bets.
At first glance this looks strategic: book profits from a mature winner (Nvidia) and redeploy into emerging opportunities in the AI space. But the size of this stake and the timing matter.
Nvidia’s valuation were elevated, to say the least, when SoftBank decided to exit. That raises the question: is this simply profit-taking, or does it reflect a shift in internal sentiment regarding where the next phase of AI growth lies?
In the institutional investor world, when a large player reduces exposure to a thematic that has carried so much of the market’s upside, it can signal two things:
1) they believe the current asset is “done” or fully valued and;
2) they may anticipate a change in leadership among winners. That in turn can influence the broader investor base.
For traders, this may be a noteworthy observation, as large institutional repositioning can sometimes influence liquidity, flows and market sentiment. What started as a small move may cascade like a waterfall if others interpret the signal similarly.
Reading Between the Lines: What These Moves Signal for AI Stocks
Taken together, Burry’s public scepticism and SoftBank’s exit from Nvidia serve as caution lights in the broader AI trade. They don’t necessarily mean the AI boom is well and truly over but they do suggest it may be time to have a reality check and “audit your optimism”.
First, both moves point toward a possible change in sentiment. When contrarians turn cautious and major institutions trim, the crowd’s belief in the narrative becomes less uniform. That can induce rotation away from the strongest winners and into second-tier names or adjacent sectors that have lagged the recent winners.
Second, we may be entering a period of market normalisation. After a run-up driven by extrapolated growth, valuations will need to be backed up by earnings growth, execution and margin expansion. That’s a tall order as the market continues to place more and more expectation on these big winners.
In the semiconductor and AI infrastructure realm that means strong end-market demand, manageable cost structures and minimal operational surprises. Any deviation, or even a slight miss, could trigger massive re-ratings.
Third, there are implications for broader indices and ETFs. The big tech-focused ETFs, like QQQ (which tracks the Nasdaq 100), and semiconductor-themed ETFs have enjoyed outsized contributions from AI-infrastructure names.
If that contribution shrinks, the momentum effect may fade. That doesn’t mean the sector collapses; it may just mean the wow factor returns to actual fundamentals rather than just the “story”.
Some market participants may monitor whether capital rotates from “pure hit” AI infrastructure names toward companies with more modest valuations, steady cash flows, and less crowded positioning.
Lessons for Traders: Sentiment, Valuations, and Market Cycles
These recent developments are a good case study in how sentiment and institutional behaviour often precede significant market shifts.
Here are five key observations framed for those following AI-related markets.
1. Understand the difference between momentum and value
The AI trade to date has been heavily momentum-driven. No surprises there. Stocks soared simply because they were “AI plays.” Moving forward, the narrative must be backed by profits (clearly from AI investments), cash flows and sustainable demand. Blindly chasing momentum increases vulnerability to sharp reversals.
2. Interpret large institutional moves
When major institutional players reposition, it may indicate their internal risk-reward calculus has changed. As in the SoftBank case, a large exit doesn’t guarantee a crash, but it signals re-examination. Use regulatory filings, reports on positions and public statements to track these behaviours.
3. Track valuation risk
When Burry raises concerns about depreciation schedules and possible earnings inflation, he’s calling attention to a hidden valuation risk. Traders must pay attention not just to revenue growth but to who is buying what, at what price and under what assumptions.
4. Plan for rotation
As we all know, markets don’t move in straight lines. At some point, the winners of one cycle become the laggards of the next. Traders who are aware of that possibility may choose to monitor for rotation rather than assume current leaders will continue outperforming. If AI infrastructure is one such cycle, what is the next one? Perhaps enterprise AI adoption, industrial AI, or regional AI deployments in Asia?
5. Diversify exposure and maintain discipline
Whether you manage trading books or direct investment portfolios, over-concentration in one theme carries higher drawdown risk. Over-investing in AI just because it “feels like the future” is a risk in itself.
Conclusion: Realism Over Optimism in AI
The recent signals from Michael Burry’s scepticism and SoftBank’s Nvidia repositioning don’t spell the end of the AI trade.
On the contrary, the underlying secular thesis of AI remains strong with compelling structural drivers but it does bear remembering that the narrative now needs to be backed up by real profits.
In that sense, when it comes to assessing the AI trade, investors and traders should both be looking at it through the lens of realism rather than fanciful optimism.
What these broader moves do signal, however, is caution and that’s a good thing; we may have moved from “growth with momentum” into “growth requiring proof.”
For traders and investors tracking AI, the message is clear to see: scrutinise valuations, heed institutional signals, be alert to sentiment shifts and avoid being caught in a crowded trade without an exit plan.
References
- “‘Big Short’ investor Michael Burry accuses AI hyperscalers of artificially boosting earnings – CNBC” https://www.cnbc.com/2025/11/11/big-short-investor-michael-burry-accuses-ai-hyperscalers-of-artificially-boosting-earnings.html Accessed 13 Nov 2025
- “SoftBank sells $5.8b Nvidia stake, sparking AI bubble concerns – TechInAsia” https://www.techinasia.com/news/softbank-sells-5-8b-nvidia-stake-sparking-ai-bubble-concern Accessed 13 Nov 2025


