Steve Eisman On Nvidia Anthropic And The AI Bubble Risk

11 min read
1 views
Nov 17, 2026

Steve Eisman, the investor who spotted the housing crash, is about to speak on Nvidia’s big report and Anthropic’s public move. What he says about the AI boom could shift the entire conversation overnight.

Financial market analysis from 17/11/2026. Market conditions may have changed since publication.

I’ve been watching the market long enough to know that when a guy like Steve Eisman decides to talk publicly about the hottest theme on Wall Street, people tend to lean in. The same investor who once saw the housing market for what it really was is scheduled to appear and discuss Nvidia, Anthropic, and the broader question hanging over everything right now: is the AI story still in full flight or starting to show cracks? That timing feels deliberate. Nvidia reports after the bell today, Anthropic is expected to step into the public markets later this week, and the conversation about whether valuations have gotten ahead of reality has grown louder by the day.

Why This Particular Conversation Matters Right Now

Markets rarely move on pure data alone. They move on narrative, and the narrative around artificial intelligence has been almost unstoppable for the better part of two years. Chipmakers, cloud providers, software platforms, and even companies with only a loose connection to the technology have enjoyed massive multiple expansion. Yet every powerful narrative eventually faces a moment of testing. Earnings season for the key players and the first real public appearance of a major frontier model company create exactly that kind of test.

Eisman’s appearance is interesting because he is not the typical technology bull. He made his name questioning excesses rather than cheering them. When someone with that track record chooses to weigh in on Nvidia’s results and the broader AI complex, the market listens for any hint of skepticism. I’ve found that these moments often reveal more about collective psychology than about the companies themselves. The questions people ask and the answers they hope to hear tell you a great deal about where sentiment sits.

Nvidia’s Earnings as the Immediate Flashpoint

All eyes are on the chip giant as it prepares to report. The company has become the purest proxy for AI spending. Every data-center build-out, every new model training run, every enterprise pilot program seems to flow through its hardware. That concentration of demand has produced extraordinary revenue growth and equally extraordinary expectations. The bar is high, and the market has shown little patience for even minor shortfalls in recent quarters.

There is an additional layer this time. Reports of potential security concerns involving China have circulated in the background. Whether those concerns ultimately prove material or not, they add a geopolitical risk premium that did not exist in earlier cycles. Investors already pricing in continued explosive growth now have to factor in the possibility of restricted sales or heightened regulatory scrutiny. That combination of sky-high expectations and new uncertainty is exactly the kind of setup that can produce sharp reactions in either direction.

In my experience, the most important part of these reports is rarely the headline numbers. Guidance, commentary on customer concentration, and any shift in tone around next-year demand usually matter more. A company that has trained the market to expect relentless upside can find itself punished for merely meeting expectations if the qualitative language feels more cautious. That is the tightrope Nvidia walks today.

Anthropic’s Public Debut and the Profitability Question

While Nvidia represents the hardware side of the equation, Anthropic stands closer to the application and model layer. The company is expected to make its public market debut later this week, and the timing is no accident. Investor appetite for pure-play AI names remains strong, yet questions about long-term profitability have grown more pointed. Training frontier models is extraordinarily expensive. Inference costs, while declining, still add up at scale. And the rapid pace of improvement across competing models raises the risk that today’s advantage becomes tomorrow’s commodity.

That last point is worth lingering on. I’ve watched several technology waves where early leaders captured enormous value only to see margins compress once the underlying capability became widely available. The same pattern could play out in generative AI. If high-quality models become easier and cheaper to access, the economic rents may shift toward companies that control distribution, data, or specialized workflows rather than the model providers themselves. Anthropic’s path to sustained profitability will be judged against that backdrop.

Public market investors are more demanding than private ones when it comes to visible cash flow. The debut will therefore serve as a real-time referendum on how much patience the market still has for growth-first AI businesses. Strong demand for the shares would signal continued confidence. A more muted reception could suggest that investors are beginning to differentiate more carefully between the various layers of the AI stack.

The Broader Question of Whether the AI Rally Is Deflating

Perhaps the most interesting aspect of Eisman’s comments will be his take on the overall AI trade. The phrase “AI bubble” has moved from the fringe into mainstream conversation. That does not automatically mean a bubble exists, of course. Every transformative technology attracts both genuine long-term capital and speculative excess. Distinguishing between the two is the hard part.

What I look for are signs of diminishing returns on incremental investment. Are companies still seeing clear productivity gains from additional AI spending, or are they beginning to question the return on the next dollar? Are enterprise customers moving from pilot projects to broad deployment, or are they pausing to reassess? The answers to those questions will determine whether the current valuation levels prove justified over a multi-year horizon.

There is also the simple math of market capitalization relative to eventual cash flows. Some of the largest AI-related companies now carry valuations that imply they will capture a very large share of future economic value. That may turn out to be correct. It may also turn out to be optimistic. History suggests the truth usually lands somewhere in between, with a few clear winners and a longer list of companies that fail to live up to peak expectations.


How Investors Are Positioning Around the Uncertainty

Walking through the market these days, you hear two distinct camps. One side argues that AI is still in the early innings and that any pullback simply creates better entry points. The other side points to stretched multiples, concentrated ownership, and the risk of a growth scare. Both views contain elements of truth. The practical question for most portfolios is how much exposure feels appropriate given the range of possible outcomes.

Some managers have reduced position sizes in the purest AI names while keeping core holdings. Others have rotated toward companies that stand to benefit from AI adoption without carrying the same valuation risk. A smaller group remains fully invested and views any weakness as noise. The diversity of approaches itself is healthy. Uniform positioning is usually the more dangerous condition.

  • Monitor guidance language more closely than headline beats or misses
  • Watch customer concentration comments for early signs of demand shifts
  • Track capital expenditure plans among large cloud and enterprise buyers
  • Compare private market valuations with public market ones for consistency
  • Keep an eye on secondary offerings and insider selling patterns

Those practical steps matter more than trying to call the exact top or bottom of the theme. Markets can remain elevated longer than most skeptics expect, and they can correct faster than most bulls prefer. Position sizing and risk management therefore take precedence over grand predictions.

Lessons From Previous Technology Cycles

It is tempting to treat AI as completely unique. In some technical respects it is. In market respects it shares traits with earlier waves. The internet build-out of the late 1990s, the smartphone and mobile internet period of the 2010s, and even the cloud computing expansion all produced periods of intense optimism followed by more selective investment. Companies that survived the first phase of excess often emerged stronger, but many others did not.

One pattern that keeps repeating is the underestimation of how quickly capabilities become standardized. Early leaders enjoy high margins while the technology remains scarce. Once competition intensifies and open-source or lower-cost alternatives appear, pricing power can erode faster than expected. That dynamic is already visible in parts of the AI landscape. Model performance gaps have narrowed in certain categories, and the cost of running capable systems continues to fall.

Another recurring theme is the lag between technological capability and widespread economic impact. The tools can be impressive long before they translate into measurable productivity gains across large organizations. Investors who focus only on the capability side risk overpaying for a future that arrives more slowly than hoped. Those who ignore the capability side risk missing structural shifts. The middle ground requires continuous reassessment rather than fixed beliefs.

What Eisman’s Perspective Adds to the Discussion

Steve Eisman built his reputation by questioning consensus when the consensus looked most certain. That history gives his comments extra weight even if one ultimately disagrees with the conclusions. He is unlikely to offer a simple “buy” or “sell” recommendation. More probably he will highlight the areas where the current narrative feels weakest and the places where genuine long-term opportunity still exists.

I expect him to address the difference between infrastructure providers and application-layer companies. Hardware companies with clear demand visibility sit in a different risk category from software businesses still searching for durable competitive advantages. He may also touch on capital intensity. Building and running advanced AI systems requires enormous ongoing investment. Returns on that capital will eventually face more scrutiny than they do today.

There is also the psychological element. When a market becomes dominated by a single theme, participants can lose perspective on valuation discipline. Eisman has seen that dynamic before. His willingness to voice discomfort, if that is his view, can serve as a useful counterweight to the prevailing optimism. Markets function better when multiple viewpoints remain audible.

Practical Considerations for Portfolio Construction

For most individual investors the goal is not to perfectly time the AI cycle. The goal is to participate in genuine long-term growth while avoiding catastrophic drawdowns. That usually means keeping position sizes reasonable, diversifying across different parts of the value chain, and maintaining some exposure to businesses that benefit from AI without depending entirely on it.

It also means accepting that volatility will remain elevated. Companies reporting strong results can still sell off if guidance disappoints. Companies reporting soft results can rally if the market decides the worst is already priced in. Emotional reactions to single data points tend to be costly. A clearer process for evaluating new information helps more than trying to predict every twist.

FactorBullish SignalCaution Signal
Demand visibilityMulti-year customer commitmentsShorter booking cycles
Margin trendsStable or expandingPressure from competition
Capital intensityClear path to returnsRising spend without clarity
Competitive positionDifferentiated technologyRapid feature parity

Using a simple framework like the one above keeps the analysis grounded. It does not eliminate uncertainty, but it reduces the chance of being swayed by pure narrative momentum.

Looking Past the Immediate Headlines

Whatever happens with today’s earnings report or this week’s public debut, the underlying technology is not going away. Computing power continues to improve. Model architectures keep evolving. Enterprise adoption, while uneven, is still progressing. The debate is less about whether AI will matter and more about how much of the current market pricing already reflects that future and how the economic value will be distributed among participants.

I’ve found that the most durable approach is to stay curious about the technology while remaining disciplined about price. Paying a fair price for a strong business usually works better over time than paying any price for a popular theme. The coming days will provide fresh data points on both the business strength and the willingness of investors to pay up for it.

Eisman’s comments will form only one piece of that data set. Still, they arrive at a moment when many portfolios have become heavily exposed to the same set of assumptions. A thoughtful critique from someone who has been right about excess before can help recalibrate those assumptions without requiring anyone to abandon the theme entirely.

The Role of Sentiment in Short-Term Moves

Short-term market moves around high-profile events often say more about positioning than about fundamentals. If a large number of investors are already long and looking for confirmation, even solid results can trigger profit-taking. If the crowd has grown more cautious, a clean beat and constructive outlook can produce a relief rally. Trying to forecast the exact reaction is less useful than understanding the range of possible outcomes and sizing positions accordingly.

One useful habit is to write down the key assumptions behind current holdings before the news hits. After the dust settles, compare those assumptions with the new information. That exercise reveals whether the thesis has strengthened, weakened, or simply moved sideways. It also reduces the temptation to invent retrospective justifications for whatever the market does.

Sentiment can shift quickly in either direction. The AI theme has already experienced several sharp corrections within the larger uptrend. Each of those corrections felt dramatic at the time and looked like buying opportunities in hindsight for those who maintained conviction. The next one may follow a similar pattern, or it may mark a more lasting change in leadership. Distinguishing the two requires ongoing attention rather than a single decisive call.

Balancing Growth Potential Against Valuation Risk

Growth investing always involves a trade-off. Higher expected growth rates justify higher multiples, but the margin of safety shrinks as those multiples expand. At extreme levels, even small disappointments in growth can produce outsized price reactions. That mathematical reality sits at the center of the current debate around AI-related stocks.

Some investors respond by focusing exclusively on the highest-quality names and accepting the valuation. Others prefer to own a broader basket at more moderate multiples. Still others look for companies that are earlier in their AI adoption journey and therefore carry less of the premium. Each approach has merits. The common thread among successful practitioners is a clear understanding of the risks they are taking and a plan for what they will do if those risks begin to materialize.

In my own work I try to separate the long-term structural case from the near-term valuation case. The structural case for increased computing intensity and software intelligence looks solid. The valuation case varies widely across individual securities. Treating every AI-related stock as identical has already proven costly for some. Differentiation will likely become more important rather than less as the cycle matures.

Final Thoughts Before the Data Arrives

The next few sessions will deliver concrete numbers and public statements from companies that have driven a large share of market returns. They will also deliver commentary from an investor who has a track record of spotting when stories have run ahead of substance. Together those inputs should help clarify whether the AI complex is still expanding on solid ground or beginning to show the early signs of fatigue.

None of this requires dramatic action. Most well-constructed portfolios already contain a mix of growth and value, cyclical and defensive, technology and non-technology exposure. The task is simply to update the probabilities assigned to different outcomes as new information arrives. That process is continuous rather than episodic.

Steve Eisman’s willingness to engage with the topic publicly is a reminder that even the strongest themes benefit from scrutiny. Markets function best when optimism and skepticism both have a voice. Today’s discussion should provide a useful dose of the latter, delivered at a moment when many participants have grown comfortable with only the former. How the market digests that conversation may prove as revealing as the earnings numbers themselves.

Whatever the immediate reaction, the longer-term questions remain the same. How durable is the demand for advanced computing infrastructure? How quickly will application-layer businesses convert impressive technology into sustainable profits? And how much of the future success is already reflected in today’s prices? Those questions will still be relevant long after this week’s headlines fade. Keeping them in focus is the surest way to navigate the noise.

I’ve watched enough cycles to know that the companies delivering real value eventually separate themselves from those riding pure momentum. The coming reports and the accompanying commentary should offer fresh evidence on which side of that line various names currently sit. Paying attention without overreacting remains the most practical stance available to most investors right now.

❝
Bitcoin is the beginning of something great: a currency without a government, something necessary and imperative.
— Nassim Nicholas Taleb
Author

Steven Soarez passionately shares his financial expertise to help everyone better understand and master investing. Contact us for collaboration opportunities or sponsored article inquiries.

Related Articles

?>