This week, the wolf of "AI will disrupt everything" has finally arrived.

The market has finally realized that AI disruption is no longer a distant threat.

On February 14th, according to ChaseWind Trading Desk, Morgan Stanley stated in its latest research report that as AI models advance non-linearly and accelerate, market pricing of disruption risks is beginning to trigger a domino effect:

Just a month ago, the market believed that about 4% of the MSCI Europe index weight was at risk of AI disruption; a week ago, this figure rose to 7%; and by February 13th, it had surged to 24% (including the banking sector).

The report notes that Morgan Stanley believes that as cutting-edge AI model capabilities reach a critical point—GPT-5.2 has already achieved or surpassed human expert levels in 71% of professional tasks—investors must reassess asset allocation logic.

Morgan Stanley has shifted its stance from neutral to cautiously viewing cyclical stocks relative to defensive stocks, highlighting that the European credit market offers cheap downside hedging opportunities, with a focus on utilities, semiconductors, defense, and tobacco, which are viewed as the most resilient safe havens.

The firm emphasizes that it is necessary to reconsider which assets cannot be “copied” by AI—these will become the new era’s value anchors. In a time when intelligence and labor can be infinitely replicated, true value will return to those things that cannot be duplicated—physical assets, regulatory barriers, network effects, human experience, and proprietary data.

AI Capability Leap: 71% of Professional Tasks Overcome

Humans are not good at understanding nonlinear changes, and AI model progress is a typical nonlinear acceleration.

Morgan Stanley states that data shows astonishing progress: Grok 4, launched in July 2025, scored 24% on the GDPVal test, meaning this model can achieve human expert levels in 24% of real professional tasks; just five months later, by December 12, 2025, GPT-5.2’s score had soared to 71%.

What is GDPVal? It’s a metric that measures AI models’ performance on real-world knowledge work, covering actual tasks performed by experienced professionals across industries. OpenAI’s research found that leading models complete these tasks about 100 times faster and at about 100 times lower cost than industry experts.

The report emphasizes that even more shocking breakthroughs are imminent. If the scaling laws for training large language models (LLMs) continue to hold in 2026—something Morgan Stanley considers highly likely—multiple state-of-the-art US LLMs are expected to be launched in the first half of 2026, with capabilities far surpassing current models. The reason is simple: the computational power used to train the top five US LLM developers is currently about ten times that of existing models.

Disruption Domino Effect: From Software to Banking

The speed of market perception change is equally astonishing.

Morgan Stanley’s tracking shows that initially, the market only questioned whether software industry revenues might sharply slow in the coming years, but this concern quickly spread like a domino to broader economic disruption risks—changing competitive landscapes, employment impacts, deflationary pressures, and more.

This recalls the market psychology evolution during early 2020’s COVID-19 pandemic: in January, it was just demand and supply chain risks; by February, it expanded to tourism, leisure, industrial, and banking sectors; by March, it evolved into a full-scale market sell-off, ultimately triggering major policy actions.

Currently, Morgan Stanley estimates that about 10% of the MSCI Europe index weight (excluding banks) is perceived by the market as facing substantial AI disruption risk; including banks, this rises to 24%. Concerns about the banking sector are relatively new, mainly centered on broader economic deflation, employment issues, and (to a lesser extent) AI-related deposit competition worries.

It’s worth noting that these “disruption stocks debated in the market” have fallen from a peak forward P/E of 24x in early 2025 to today’s 16.4x. But Morgan Stanley warns that if we look at the valuation trend of those “undisputed disruption stocks”—which fell from 24.7x to 11.1x—there may still be further downside potential.

Who Can Survive in the AI Era?

Faced with this disruption storm, Morgan Stanley offers a framework combining five dimensions to assess sector and individual stock resilience:

AI Exposure Level: Whether a stock is a disruptor, a “disruption debate target,” an enabler, or protected

Business Nature: Service provider, physical assets, commodities, or compute power

Cyclicality: Cyclical stocks, defensive stocks, or others

Investor Positioning: Current holdings level

Stock Momentum: Fundamental factors plus additional signals

Based on this framework, Morgan Stanley considers the most resilient sectors to be: utilities, semiconductors, defense, tobacco, and personal & household care.

Morgan reports that European utilities nearly dominate the top 20 most disruption-resistant stocks. These companies share common traits: providing physical infrastructure that AI cannot replicate, being defensive in nature, and currently underweighted in portfolios.

Conversely, software, business services, media & entertainment, travel & leisure, as well as transportation, diversified financials, and banks, are seen as facing the highest spreading disruption risks.

Eight Asset Classes That Cannot Be Copied by AI

Meanwhile, Morgan Stanley emphasizes that once AI reaches a transformative level, the value of those assets that cannot be “copied” by AI will rise. This is a key framework for understanding future asset allocation:

A. Physical Scarcity: Real estate, energy and power assets, transportation infrastructure, data centers, mineral metals, water resources, casino licenses in limited jurisdictions, theme park land, cruise port and terminal rights, spectrum licenses, fiber optic networks, etc.

B. AI Adopters with Pricing Power: Capable of demonstrating increasing pricing power thresholds.

C. Unique luxury goods, real estate, and services.

D. Network Effects: Large tech platforms, online marketplaces, healthcare businesses with patient relationships.

E. Truly Unique Human Experiences: Media brands, sports assets/teams, music and performance arts that emphasize human elements.

F. Regulatory Scarcity: Businesses with various licenses, approvals, and protected rights.

G. Proprietary Data and Brands: AI adopters with proprietary datasets and IP portfolios.

H. A Range of Semiconductor Assets: Leading process technologies, ASML’s EUV lithography, TSMC’s manufacturing expertise, rare chip materials processing.

Credit Market: Cheap Downside Protection

Although AI disruption fears have begun to impact some credit markets, especially leveraged loans, European investment-grade spreads remain near post-2008 lows. Even as implied equity volatility rises, credit volatility remains unusually low.

However, if AI disruption fears spread to more sectors (along with the expected issuance acceleration), it could challenge the resilience of the credit market.

Morgan Stanley believes that credit options markets offer good entry points for investors to prepare for widening spreads. Given Europe’s relatively low tech exposure, overall yields still high, policy support, and economic resilience, these hedging tools are especially cost-effective.

Computing Power Gap: An Invisible Supply Crisis

On the other side of AI disruption is the frantic demand for compute infrastructure. Multiple data points show that demand growth far exceeds current supply forecasts:

• Google executives recently stated that the company may need to double compute capacity every six months—“reaching 1000x in 4-5 years.” In comparison, Morgan Stanley forecasts a compound annual growth rate of about 210% for Nvidia’s compute sales from 2025 to 2028; over five years, this would amount to roughly 300x, far below Google’s required 1000x.

• OpenRouter data shows that from late November 2024 to late November 2025, weekly average token demand increased by over 2200%. Token usage is a direct proxy for compute demand.

• More critically, the computational intensity of individual LLM queries is rising rapidly. Research firm METR notes that the average “work” per customer query doubles every 7 months.

According to the report, even with a stable customer base, this growth means compute demand will outpace Nvidia’s estimated 120% annual growth rate by a significant margin.

Morgan Stanley states that this supply-demand imbalance is already evident in the market:

CoreWeave can lease older Nvidia Hopper GPUs at 95% of original price, well above the depreciation implied by chip economics over time;

The power supply lease deal guaranteed to Anthropic and FluidStack provides Bitcoin miner Hut8 with about 18.5% unleveraged capital return, equivalent to a power access premium of roughly 300%.

All the above insights are from ChaseWind Trading Desk.

For more detailed analysis, real-time insights, and frontline research, join the **ChaseWind Trading Desk Annual Membership**.

![](https://img-cdn.gateio.im/social/moments-4b104798ea-2969ab572c-8b7abd-d8d215)

Risk Disclaimer and Terms of Liability

Market risks are inherent; investment decisions should be made cautiously. This article does not constitute personal investment advice and does not consider individual user’s specific investment goals, financial situation, or needs. Users should consider whether any opinions, viewpoints, or conclusions herein are suitable for their particular circumstances. Investment involves risk, and responsibility rests with the individual investor.