The AI trade was pushed into a ditch this week.

Anthropic CEO Dario Amodei’s weekend warning about the pace of AI development triggered a sharp reassessment of the AI infrastructure narrative. The semiconductor complex absorbed the initial shock, with the PHLX Semiconductor Index experiencing a significant sell-off at Monday’s open.

The market’s reaction was understandable.

The AI trade has increasingly become a capital-expenditure trade. The spending plans of Microsoft, Alphabet, Amazon, Meta, and other hyperscalers have helped support the broader AI ecosystem, from advanced semiconductors and data centers to networking, power, and cooling infrastructure.

As a result, any suggestion that AI development could slow—even for safety or regulatory reasons—immediately raises a more consequential question:

What happens if the capital-expenditure cycle loses momentum?

The market is not simply pricing AI technology. It is pricing the continuation of an enormous investment cycle.

Meanwhile, the Federal Reserve raised its benchmark interest rate by 25 basis points on Wednesday. Fed Chair Kevin Warsh’s hawkish messaging initially pressured risk assets, while subsequent comments from President Trump on Truth Social coincided with a reversal in market sentiment.

Two major themes dominated the week:

The market is entering a period where sector selection may matter more than simply maintaining broad exposure to risk assets.

The answer depends on which layer of AI we are discussing.

Nvidia CEO Jensen Huang has described the AI ecosystem through five layers:

These layers are connected, but they do not necessarily have identical growth rates, economics, or investment cycles.

Amodei’s warning primarily challenges the assumptions surrounding the pace of frontier AI development.

The market, however, immediately translated the message into a financial question:

Could the AI infrastructure spending cycle slow down?

That is the critical distinction.

A discussion about safety, model development, or regulatory oversight does not automatically mean that demand for AI applications will decline. But it can cause investors to reassess the pace, timing, and return on investment of the infrastructure being built to support those applications.

The five layers may therefore experience different outcomes.

AI data centers require substantial and reliable electricity.

Even if the pace of model development moderates, existing demand for power infrastructure, grid investment, and energy capacity may remain significant. The timing of new projects, however, could be affected if hyperscalers become more cautious about capital allocation.

This is where the market reaction was most immediate.

Semiconductor valuations are closely linked to expectations for future AI computing demand, advanced accelerator shipments, networking requirements, and pricing power.

A slowdown in the rate of infrastructure expansion would not necessarily eliminate long-term demand. It could, however, affect order growth, inventory expectations, capital-expenditure forecasts, and valuation multiples.

The market may increasingly distinguish between:

Cloud providers and data-center operators face a more complicated equation.

They must invest ahead of demand while managing financing costs, power availability, utilization rates, and customer returns.

The question is no longer simply whether AI demand exists.

It is whether the incremental return on every additional dollar of infrastructure investment remains attractive.

Foundation-model developers are at the center of the debate.

They are responsible for pushing model capabilities forward, but they also face significant costs associated with computing, research, safety, talent, and deployment.

A more deliberate development pace could alter the timing of capital expenditures. At the same time, improved efficiency, better model architectures, and more effective training techniques could eventually reduce the cost of delivering AI capabilities.

A slower pace of development is not automatically equivalent to a smaller long-term market.

It may instead mean that the market must revise its assumptions about how quickly capital converts into revenue and earnings.

This is where our view becomes more constructive.

We believe the application layer could continue accelerating in the coming quarters, particularly as companies gradually determine how to deploy AI agents and automation tools to improve productivity, reduce expenses, and redesign workflows.

Many businesses are still in the experimentation phase.

They are testing how AI can be integrated into:

The commercial opportunity may expand even if the pace of frontier model development becomes more measured.

The infrastructure cycle and the application cycle are related—but they are not the same trade.

Dario Amodei’s comments arrive at a time when big tech’s capital expenditure has already been under relentless scrutiny.

The spending cycle largely carries the AI trade. If investors begin to believe that hyperscalers are preparing to slow investment, the market could experience a significant repricing across semiconductors, data centers, networking, and related infrastructure.

From that perspective, Amodei’s warning may ultimately serve a useful purpose for the broader AI ecosystem.

It introduces the possibility of a slowdown into investor expectations before the market is forced to confront one through disappointing earnings or reduced guidance.

That does not eliminate the risk of a sell-off. It may, however, help investors distinguish between:

Our view is that the application layer of AI could continue to accelerate in the coming quarters, particularly as enterprises move beyond experimentation and begin deploying AI agents to improve operating efficiency.

Healthcare, financial services, and selected technology companies may benefit from this next stage of adoption.

There is also a more nuanced potential benefit.

As enterprises deploy AI systems, their successes and failures can generate valuable feedback for the major large-language-model developers. That feedback may be delivered voluntarily through collaboration or involuntarily through real-world deployment outcomes.

Either way, the application layer becomes an important testing ground for the next generation of AI development.

The companies that learn how to deploy AI effectively may become the bridge between model capabilities and measurable economic productivity.

As highlighted in our previous newsletter, navigating the inflation and interest-rate debate remains a delicate balancing act between three major policy forces:

These institutions do not have identical mandates.

Their objectives may overlap, but they can also come into conflict.

The Fed may be concerned about inflation persistence.

Treasury must manage substantial borrowing requirements and the functioning of the long-term bond market.

The administration has emphasized growth, domestic investment, and maintaining U.S. leadership in AI.

The market’s challenge is to determine how these competing objectives translate into the future path of the 10-year Treasury yield.

The sequence of events was revealing.

The Fed delivered a 25-basis-point rate hike, while Warsh’s messaging reinforced concerns about inflation and borrowing costs. The 10-year Treasury yield has remained under pressure, with the cost of capital increasingly influencing equity valuations and investment expectations.

The subsequent shift in market sentiment following Trump’s Truth Social post demonstrated how sensitive markets have become to policy communication.

We should be careful not to conclude that a single post can sustainably control the Treasury market.

The bond market ultimately responds to a much broader combination of:

However, the market’s reaction reinforces the point made in our previous newsletter:

The 10-year Treasury yield is the place where the policy debate ultimately becomes a market price.

One important consideration is the relationship between earnings growth and the cost of capital.

If corporate earnings are projected to grow at rates in the low 30% range, as discussed in our market framework, that could provide companies and investors with a substantial earnings-growth cushion against temporary interest-rate shocks.

This does not guarantee that equities will rise.

Valuations still matter, and higher yields can compress the multiple investors are willing to pay for future earnings.

But strong earnings growth may give policymakers more room to respond to short-term economic shocks without immediately destroying the broader growth narrative.

The situation becomes more complicated if:

Conversely, if earnings remain strong while inflation moderates, the Fed and Treasury may have greater flexibility to reduce the intensity of their messaging or respond to market stress.

A softer projected PCE inflation path in 2027 would be relevant to that discussion, although future inflation projections remain uncertain and should not be treated as guaranteed outcomes.

The broader point is this:

The Fed, Treasury, and administration may have more room to manage short-term shocks than the market initially assumes—but that flexibility depends on inflation expectations, bond-market confidence, and the resilience of corporate earnings.

The ballroom may be large.

The challenge is whether the participants can coordinate their movements without unsettling the market.

The market is entering a high-growth, elevated-rate, and inflation-sensitive environment.

That combination favors a more differentiated approach to sector exposure.

Rather than treating the market as a single risk-on or risk-off trade, investors may want to examine which industries have the strongest structural growth drivers, which are most sensitive to long-term yields, and which have relatively different macroeconomic exposures.

Information Technology remains closely tied to the AI investment cycle, digital transformation, enterprise software, and productivity improvements.

Even after the recent semiconductor sell-off, the sector’s long-term growth drivers remain relevant.

However, investors may need to distinguish between the different parts of technology:

The next phase may not reward every company equally.

Companies demonstrating measurable revenue growth, operating leverage, and customer adoption could be evaluated differently from companies whose outlook depends primarily on continued expansion in capital spending.

Healthcare, biotechnology, and genomics deserve increased attention as AI becomes more deeply integrated into research, diagnostics, drug discovery, clinical workflows, and healthcare administration.

This could become one of the most important long-term applications of AI.

Potential areas of transformation include:

The economic impact may take time to develop, and the sector carries its own risks, including clinical failure, regulatory uncertainty, reimbursement constraints, and financing requirements.

Nevertheless, healthcare and genomics could become important structural growth themes as AI capabilities improve.

For investors considering long-term portfolio construction, the sector merits attention not only as a defensive allocation but also as a potential beneficiary of technological transformation.

Financial services may benefit from the continued adoption of AI in research, fraud detection, customer service, underwriting, compliance, and operational automation.

Asset managers, exchanges, brokerages, financial technology platforms, and other market-infrastructure businesses may have different exposure profiles from traditional banks.

The key question is whether AI adoption translates into:

Financial services remain sensitive to credit conditions, the yield curve, and economic growth. Sector exposure should therefore be evaluated alongside the broader interest-rate environment.

Crypto’s Friday move, despite negative developments surrounding the Clarity Bill, is worth monitoring.

The market’s ability to rally against a negative legislative backdrop may suggest that crypto remains in a broader cyclical bull-market structure. That interpretation should be tested against price trends, liquidity conditions, regulatory developments, and institutional demand.

The legislation front remains a potential catalyst.

Positive developments could produce a sharp move as investors reassess the regulatory outlook and the potential for broader institutional participation.

However, crypto remains highly sensitive to:

The broader thesis is not simply that every regulatory development will be positive.

It is that a constructive regulatory surprise could generate a disproportionately large market response when positioning and expectations are already established.

Agriculture and fertilizer remain among the sectors with relatively different exposures to the Fed and geopolitical developments.

The sector is not immune to wars, trade restrictions, energy prices, or currency fluctuations. However, its fundamental drivers include food security, crop yields, global supply conditions, agricultural productivity, and input costs.

The sector appears to be in a bullish phase based on the market behavior we have been tracking.

The more important question is whether this represents a short-term momentum trade or the early stage of a broader multi-year cycle.

Potential structural drivers include:

Compared with crowded technology trades, agriculture and fertilizer may offer a different source of exposure to global economic and supply-chain trends.

This week’s market action was also influenced by the largest options expiration cycle of the quarter.

Large expiration events can increase short-term trading activity, hedging adjustments, and mechanical flows across indexes and individual securities.

It is difficult to attribute every market move to options positioning alone.

Nevertheless, when markets are already operating near a technical or macroeconomic pressure point, large derivative-related flows can amplify volatility.

The market appears to be accumulating pressure.

The question is what happens when that pressure is released.

Our interpretation is that the next directional move could be significant, particularly if the market receives a catalyst that changes expectations around:

The direction should not be treated as predetermined.

But the combination of crowded positioning, elevated macro sensitivity, and large options expiration creates an environment where a breakout or breakdown could generate meaningful momentum.

Last week’s market action reinforced several themes we have been following.

The semiconductor sell-off demonstrated how vulnerable the infrastructure layer can be to changes in expectations surrounding AI development and capital expenditure.

But the application layer may follow a different trajectory.

Enterprises are still working to determine how AI agents can improve productivity, reduce expenses, and transform their operations.

That creates a potential opportunity beyond the infrastructure buildout.

The next phase of AI may be defined less by how much companies spend and more by how effectively they convert AI investment into economic output.

The Fed’s decision, Treasury’s financing requirements, and the administration’s growth objectives all converge in the bond market.

The 10-year yield remains a critical signal for equity valuations, financial conditions, and the sustainability of risk-on positioning.

If yields stabilize while earnings remain resilient, growth-oriented sectors may have room to recover.

If yields accelerate higher while earnings expectations deteriorate, the market could experience another broad repricing.

Our areas of focus heading into Q4 2026 include:

These themes carry different risks, time horizons, and valuation considerations. They should not be treated as interchangeable exposures.

The combination of AI uncertainty, elevated Treasury yields, Fed policy, geopolitical risk, and large options expiration has created a market under pressure.

Once that pressure is released, the resulting move could be significant.

The next catalyst may come from the bond market, the Fed, corporate earnings, AI capital-expenditure guidance, or regulatory developments.

The key is to remain attentive to the market’s reaction rather than relying solely on the headline.

The pressure is building. The question is not whether momentum will return, but which sectors will capture it when it does.