The Third Wave of AI: Why It’s Time for Investors to Pivot From ‘Infrastructure’ to ‘Efficiency’

Investors have likely noticed a recurring pattern recently: The moment the slightest negative news hits the market, the semiconductor sector and U.S. indices instantly go into a deep dive. First, we saw sharp selloffs on news from Broadcom (AVGO) and SpaceX (SPCX). Most recent was a staggering 10% collapse in South Korea’s KOSPI Index ($KSIC), which ricocheted and hit the Nasdaq ($NASX). In my opinion, this hypersensitivity is a clear diagnosis. The market is overextended to the limit, sitting on a powder keg of margin positions, and major players now need literally any excuse — no matter how insignificant — to lock in profits.

These dramatic cracks in the global semiconductor complex are more than just a technical correction. In my view, they represent a fundamental shift in the market regime. For the past two years, investing in artificial intelligence (AI) has been a straightforward, one-way bet. Investors bought the companies building the hardware. Today, this trade seems to have reached its limits.

The risk-reward ratio in the semiconductor and hardware infrastructure sectors has inverted. In my opinion, while the structural demand for AI chips will remain high through 2026, the potential for further outsized gains is now more than offset by mounting downside risks. Every marginal dollar invested in overheated chipmakers at current prices is no longer a long-term investment, but a high-stakes speculation on an increasingly narrow tightrope.

To survive the next phase of the market cycle, I recommend investors change course. I believe the era of buying the "builders" of AI infrastructure is giving way to an era of buying its “implementers” — companies in the real economy that are turning AI technology into hard, bottom-line efficiency.

The Evolution of the AI Investment Lifecycle

To understand where I believe the "smart money" is moving now, we must break down the AI revolution into logical, chronological waves.

The first wave had to do with software and hyperscalers. The boom began with the direct creators of large language models (LLMs) and software platforms. Mega-caps like Microsoft (MSFT) led the charge. However, over time, this wave lost momentum, shifting into a multi-month sideways consolidation.

Next came the infrastructure gold rush. Realizing that AI requires unprecedented computational power, investors poured capital into physical infrastructure. This was the most explosive phase of the cycle. First, Nvidia (NVDA) went vertical, followed by a domino effect across the entire supply chain — from memory makers like Micron (MU) and SK Hynix to server integration and cooling specialists like Supermicro (SMCI) and Vertiv (VRT) to energy-grid component providers like GE Vernova (GEV).

Today, this second wave seems to be experiencing severe structural fatigue. When the slightest rumor about memory deployment schedules or a geopolitical shift can trigger a double-digit collapse of an industry giant, in my view, that is a clear sign of an overbought market. The hardware revenue of these companies is largely locked in for 2026 by existing corporate budgets, but stock prices have already pulled forward all possible future optimism. I believe that hunting for "laggards" in the semiconductor sector has now become a trap, as the margin of safety has evaporated.

With that said, I think it is time to look for other companies. The third wave will be an era of corporate optimization. The "infrastructure tax" has already been paid to the tech giants. The data centers are built, the chips are installed, the cables are laid. The main question now is shifting from "Who is selling the tools?" to "Who is using these tools to maximize profits?"

Rethinking Returns: From 5x Multipliers to Margin Expansion

Investors must radically adjust their expectations. The speculative mania of the second AI wave has conditioned market participants to expect 500% returns in 12 months. But I think the third wave will look completely different.

If a classic blue-chip company like JPMorgan Chase (JPM), Walmart (WMT), or UnitedHealth (UNH) successfully deploys autonomous AI agents to optimize its workflows, its stock won't trade like a hyper-growth tech startup. Shares won't double overnight. Instead, in my opinion, the third wave of AI will be a margin expansion trade.

I expect the winners of this phase to be traditional, highly liquid enterprises using AI to execute quiet, internal revolutions. This will include aggressive de-risking and operating expense reduction, accelerated research and development and time-to-market, and the optimization of physical assets:

• Aggressive De-risking and Operating Expense Reduction: Companies capable of automating heavy back-office operations, compliance, auditing, and basic customer support will drastically cut their operating expenses. I believe that for a financial institution or insurance giant, a 30% reduction in administrative costs can translate directly into billions of dollars in pure profit.
• Accelerated R&D and Time-to-Market: In sectors like pharmaceuticals and biotech, AI is shortening drug-discovery timelines. The capital efficiency achieved by avoiding failed chemical trials is colossal.
• Optimization of Physical Assets: In heavy industry, agriculture, and logistics (think Caterpillar (CAT) or Deere (DE)), AI is also being implemented for predictive maintenance and autonomous operations. Eliminating unscheduled equipment downtime or cutting chemical fertilizer costs significantly would provide a permanent structural advantage over less advanced competitors.

A New Investment Strategy: Follow the Buyers

As the market turns away from overbought tech names, institutional managers are looking for companies where AI is viewed as a core architectural upgrade, not a marketing gimmick. Future corporate earnings reports will help separate the market into two distinct groups — those buying AI for flashy headlines and those buying it for systemic cost reduction.

In my view, future premium valuations will go to companies that can demonstrate a clear divergence between revenue growth and headcount expansion. If an enterprise can scale its business by 20% while keeping administrative and general expenses flat thanks to AI productivity, it would likely become an incredibly high-quality cash-generating machine.

Conclusion

Gradually, the value of the AI revolution will move down the value chain to the end consumer. For long-term investors, this represents a return to fundamental, earnings-based investing.

It's time to stop looking for the next chipmaker. Instead, investors should start looking for traditional, established businesses that are quietly adopting AI to gut their cost structures, protect their margins, and capture market share. I believe this is exactly where the real wealth of the next market cycle will be forged.