Author: Lone Ranger
Recently, the bull and bear signals in the AI sector have crowded the market, intensifying concerns among investors.
Goldman Sachs recently released its 2027 global market outlook report, indicating that the accelerated wave of AI capital expenditure is likely to peak in early 2027, transitioning from a unidirectional upward trend to a two-way volatility, and suggesting an increase in diversified allocation and hedging;
The Federal Reserve also clearly called out artificial intelligence in its latest financial stability report, warning of its potential threats to asset valuation and financial stability.
Notably, well-known bear Michael Burry has opted to express his stance directly through positions, doubling his put options on the NASDAQ 100 while re-establishing short positions insemiconductor ETFs, stating that this is a "bull market trap," particularly benchmarked against the dot-com bubble peak of 2000. Jensen Huang's departure has further heightened his concerns about the entire AI industry chain.
01
Specifically, several reports point to several hard constraints facing the current AI narrative:capital expenditure, high interest rates, Democratic regulatory constraints, and concerns about the trustworthiness of the supply chain.
These factors are also the sources of recent market worries.
The Goldman Sachs report states that the ratio of AI investment increment to GDP will peak in early 2027, but it does not imply that capital expenditure will turn downward; rather, the pace of increase in investment is slowing.
The stock market's pricing typically places greater emphasis on marginal changes. The previous bull market in the upstream computing power sector relied on continuously rising capital expenditure growth. Once this momentum stops, the core support for high valuations will loosen.
According to Goldman Sachs' estimates, the capital expenditure of cloud vendors has a clear correlation with the pull on S&P 500 EPS, contributing 2 percentage points in 2024, 5 points in 2025, peaking at 12 points in 2026, dropping to 9 points in 2027, and turning into a 5-point drag on earnings by 2028.
By 2026, it is basically the last window for capital expenditure to drive earnings.
The decline in the growth rate of earnings also reflects this change, with S&P related earnings expectations dropping from 36% to 11%, and European, Japanese, and Asian markets also weakening simultaneously.
Objectively speaking, an 11% growth rate is not bad historically. The discomfort stems from the drop from 36% to 11% during this period.
Previously, the market enjoyed the dual boost of rising earnings and valuations, but moving forward, it is likely that only slow earnings support will remain, and valuations will continue to be under pressure, making it difficult to replicate the previous two years of easy profits.
The second unavoidable reality constraint is the high interest rate market environment.
The suppression of the AI sector by high interest rate environments is twofold.
AI is a typical long-duration asset, with most of the returns expected in the long term. An increase in long-term interest rates directly raises the discount rate, putting downward pressure on valuations.
Major cloud players like Microsoft and Google have strong foundations and abundant cash flows, experiencing limited impacts; the most affected are the mid-tier GPU cloud service providers, such as CoreWeave and Nebius, which Burry named. They rely on large-scale leasing and purchasing of GPUs for rapid expansion and are highly dependent on external financing.
When interest rates rise, their borrowing costs increase directly, forcing contractions in expansion. Moreover, their valuations were based on extremely high growth rates, making the expectation gap easy to widen.
The cyclical trade risks mentioned by the Federal Reserve are also worth noting.
Tech companies and startups mutually invest, procure, and endorse each other's revenue. When funds are loose, it can beautify financial statements and inflate valuations. After financing tightens, it turns into a series of linked risk exposures; if one company faces cash flow issues, the risk will propagate along the cooperation chain. This is not just a matter of valuation; it is a potential credit risk.
The biggest recent negative trigger has been the Jensen Huang incident, which has shaken the pricing anchor of the entire chain.
It is still difficult to conclude how much long-term impact this event will have on Nvidia's products and orders.
In the past few years, the orders and valuation logic for many companies in chips, foundries, storage, and downstream cloud vendors have been built around Nvidia's strong position. Once this anchor shakes, everyone will reassess their assets.
The big bear Burry has clearly seized this window, but his thinking has also changed.
Previously, he favored directly betting on Nvidia andPalantir, but now he has shifted focus to indices and semiconductor ETFs. He has simultaneously doubled his put options on the NASDAQ 100 while buying back short positions in the VanEck semiconductor ETF, where TSMC and Nvidia are heavy holdings.
He has aligned the recent peak of the NASDAQ in early October with the 2000 bubble peak, declaring a "bull market trap," specifically watching whether Nvidia's influence will gradually weaken.
Of course, his strategy may not be right, as the comparison is significantly different.
During the 2000 bubble, many companies had no cash flow and were riddled with debt; now, the leading tech giants in US stocks have relatively healthy balance sheets, and the valuation premium of the top five tech stocks compared to other stocks has contracted to a very low level, far from the frenzy of previous years.
Additionally, this options bet has a high threshold, requiring the index to drop over 20% before expiration to be profitable, making it very easy to falter on time, magnitude, or direction.
Another layer of pressure that is easily overlooked comes from the regulatory shift.
After the Democrats came to power, the US's attitude towards AI infrastructure will likely gradually shift towards regulation; the energy consumption of computing clusters and large data centers, land use, and electricity supply are all starting to face more constraints.
This doesn't mean AI will be stifled, but it will disrupt the rhythm of uncontrolled computing power growth, elongating expansion cycles and increasing construction costs.
Looking back at the capital expenditure curve, this variable will further lower the expansion slope. Expectations of sustained explosive growth in computing power will be challenged by increasing approvals, energy consumption, and regulatory constraints, with peaks arriving earlier and the pace of realization slowing.
With four layers of pressure stacking up, it is only natural for the valuation bubble inflated by optimistic narratives to be squeezed.
02
However, we must be clear that the collapse of AI's valuation bubble does not mean the AI industry is on a downhill path; the underlying demand supporting AI has not disappeared.
There is also an interesting signal that Goldman Sachs has mentioned multiple times.
The forward PE of US cloud giants is only 19.7 times, below its own historical lower bound of 20 years, and even lower than the widely recognized defensive sector of consumer staples.
Many people view high elasticity growth stocks, as their valuations have fallen below those of defensive stocks, indicating that a significant portion of pessimistic expectations is already factored into the price, and not everyone is blindly optimistic.
Long-term regulatory tightening is also not entirely a bad thing. The chaotic arms race for computing power will taper off, forcing everyone to calculate whether the invested computing power can yield stable revenues and profits.
Only with viable business models that do not rely on endless cash burning to expand capacity will the AI narrative become more solid. In the short term, it may be painful, but in the long term, it is about filtering out the false to retain the true.
After the bubble has deflated, the market's opportunity structure will also change to a new logic.
The simplest strategy in the past two years was to go all in on upstream mining stocks—chips, optical modules, storage, where the closest to computing power saw the best gains. This logic is based on the sustained acceleration of capital expenditure and will be difficult to replicate going forward.
The market's assessment criteria will shift from "who is expanding production" to "who can make real money."
The first category worth tracking is the cloud giants whose valuations have adjusted appropriately.Slower capital expenditure growth may actually mean less investment and improved free cash flow for them. With solid balance sheets and diversified revenue structures, they are slowly transforming from high-volatility growth to defensive varieties with growth attributes, where cost-effectiveness will gradually become more apparent in turbulent markets.
The second category is the AI application companies that have truly taken off.In the past, the market's focus was entirely on computing power, overshadowing the value on the application side.
Moving forward, there will be repeated questioning of whether AI can help clients save money and increase revenue, and whether there are stable payments and contract renewals. Only those companies that can realize genuine productivity gains and have clear ROI may take the lead as a new mainstay.
This corresponds to what Goldman Sachs refers to as the "successor team." How smoothly this transition goes will determine whether this round of adjustments results in a soft landing or a continual downturn.
The third category consists of cyclical sectors that have already priced in extensive pessimistic expectations, such as South Korean memory stocks.Stock prices already imply expectations of a drastic 70% reduction in earnings by 2027, yet orders are scheduled through 2029, with PB ratios compressed to historical lows. While the odds appear favorable, the risks are not insignificant, with extreme volatility.
Goldman Sachs has also not blindly turned bullish on Asia, with core concerns centered on excessive concentration; only a few targets like TSMC, Samsung, and SK Hynix account for a heavy proportion, meaning that if the logic reverses, the pullback could be very severe.
Conversely, there are also many targets that need to be actively avoided. Mid-sized cloud vendors heavily tied to Nvidia orders, lacking stable cash flow and relying on financing, as well as purely cash-burning startups without viable products, represent the weakest groups in this pressure test.
03
Overall, the current AI narrative is indeed facing tremendous pressure, and in this market context, holding onto the previous years' strategy of heavily investing in upstream computing power and continuing to pay for "faster and faster growth" easily leads to losses.
A more prudent approach is to reduce concentrated bets on single lanes and moderately diversify positions.
The market has now shifted from a unidirectional bull market to two-way fluctuations; thus, the value of hedging and diversified asset allocation will be higher than in previous years.
This round of testing for the AI narrative does not challenge whether AI can change the world but whether these investments can translate into stable returns. The process of deflating the bubble is bound to be difficult, with volatility continuing for a long time, but after the market clears, assets that can withstand scrutiny and have solid cash flows will usher in the next leg of a longer-lasting trend.
During this time, it is essential to maintain composure under pressure and patiently await the dawn.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到[email protected]，本平台相关工作人员将会进行核查。
