Boston Fed President Susan Collins, a voting member of the FOMC in 2025, stated clearly on Wednesday that there is “a relatively high bar” for further policy easing in the near term. She noted that with inflation still above the Fed’s 2% target and the government shutdown causing gaps in economic data, keeping rates unchanged is the appropriate choice.

On Thursday, San Francisco Fed President Mary Daly echoed that stance, saying, “It is too early to conclude that we definitely won’t cut rates, or that we definitely have to cut.” She stressed that data over the next four weeks will be critical, while also acknowledging that the labor market “has clearly slowed.”

Fed's Collins expresses hope that inflation can be tamed without hitting jobs

(Image source: Internet)

Rifts inside the Fed continue to widen. At the October meeting, Governor Adriana Kugler Miran (Miran in the text) argued for a 50-basis-point rate cut, while Kansas City Fed President Jeffrey Schmid opposed any cuts at all. This divergence has sent the probability of a December rate cut plunging from 70% last week to 51.9%, triggering a sharp market reaction.

On Thursday, all three major U.S. stock indices closed lower: the Nasdaq tumbled 2.29%, the S&P 500 fell 1.66%, and the Dow lost 1.65%. Mega-cap tech stocks were broadly sold off, with Nvidia down 3.58%, Microsoft off 1.54%, and Tesla plunging 6.64%.

Microsoft Rejects the “Compute Bet,” Redefining AI Infrastructure Logic

In a deep-dive interview on November 13, Microsoft CEO Satya Nadella unveiled a major strategic adjustment: the company will no longer pin its fate on providing low-margin compute infrastructure for a handful of AI model companies. Instead, it will moderate the pace of capacity expansion to align with Nvidia’s technology upgrade cycle.

Nadella was blunt: “I don’t want to be trapped by the sheer scale of one generation of technology. Pace matters; substitutability and location matter; workload diversity matters; customer diversity matters.”

This shift is designed to avoid being “locked in” by the four-to-five-year depreciation cycle of a single generation of hardware, directly addressing the depreciation risks highlighted by “Big Short” investor Michael Burry.

Microsoft is moving away from “bare-metal services for a few customers” toward building systems that support a diverse range of AI workloads. Nadella emphasized, “We are in the hyperscale business, and essentially it’s a long-tail business for AI workloads,” signaling that Microsoft aims to serve thousands of enterprises with different AI needs, rather than just a few large model players.

At the core of this strategy is boosting capital efficiency. Nadella noted, “We are now both a capital-intensive and knowledge-intensive company. We must use our knowledge to improve the return on capital expenditure.” Microsoft is seeking to maximize hardware utilization through software optimization, turning spending into competitive advantage.

Focus on Data and AI Agents

Nadella offered a sharp insight: AI models themselves face the risk of commoditization. “If you are a model company, you might fall into a ‘winner’s curse’,” he said. “You’ve done all the hard work and achieved incredible innovation—but it only takes one act of copying for it all to become commoditized.”

The open-source community is accelerating that process. Nadella pointed out, “Structurally, there will always be a fairly capable open-source model in the world, as long as you have something to use it with—namely data and scaffolding.”

This implies that long-term value may not lie at the model layer, but in the application ecosystem built on top of it. Based on this view, Microsoft is shifting its focus from model development to the data and application layers. Nadella believes that future value will reside “with the people who own the data used for foundation and context engineering” and with AI agents capable of executing complex tasks.

Microsoft’s product strategy is evolving accordingly—deeply embedding AI into its product suite instead of merely adding a chat layer on top. The company is building infrastructure that supports a wide variety of workloads from pre-training to inference, going beyond generative AI, to ensure it remains indispensable even as models themselves become more commoditized.

Tech Stocks Face a Repricing

The recent pullback in tech stocks reflects a dual concern in the market. On one hand, the Fed’s hawkish tilt is weighing on liquidity-dependent assets; on the other, the risk of over-investment in AI infrastructure is drawing increased scrutiny.

Charles Schwab’s chief macro strategist commented, “Markets with high valuations tend to struggle when anxiety spreads—that’s part of the digestion process.” As policy uncertainty at the Fed persists, the valuation framework for tech stocks is coming under pressure.

Microsoft’s strategic pivot provides an important reference point for the entire tech sector. By refusing the “compute bet,” the company is confronting a core contradiction in the AI boom: in an environment of rapid technological iteration, blind investment in hardware can lead to massive depreciation risks.

Microsoft’s share price fell 1.54% on Thursday, but its medium- to long-term performance will depend on whether the new strategy can strike the right balance between capital efficiency and technological leadership. This shift may also reshape the broader AI value chain, nudging it away from a simple race for compute scale toward a more refined, efficiency-driven growth model.