There’s a fight happening right now over the future of artificial intelligence, between a White House that wants no guardrails and an AI company whose Pentagon contract is under review for insisting on them, between companies spending hundreds of billions on a productivity revolution and a market that’s rotating away from them in real time. Who profits, who gets displaced, who sets the rules, these questions are being answered right now, not in theory but in stock prices, in Washington, and in the portfolios of anyone with meaningful exposure to U.S. technology.
But first, the numbers.
A Rotation Hiding in Plain Sight
The year started on an upbeat note for stock investors. U.S. and global equities posted January gains, bonds delivered modest returns, and the Federal Reserve held interest rates steady at 3.5%–3.75% after cutting three times in late 2025. The S&P 500 gained 1.5% for the month. The Russell 2000 returned 5.35%. International developed markets gained nearly 5%, and emerging markets surged almost 9%.
But the composition of that return was anything but ordinary. Small-cap stocks outperformed large-caps. Value stocks outpaced growth across every capitalization band. And international stocks, both developed and emerging, handily beat U.S. equities.
In the U.S., large-cap value stocks returned 4.56% for the month while large-cap growth stocks lost 1.51%, a spread of more than six percentage points in a single month. U.S. small-cap value gained 6.86%. Non-U.S. developed markets rose nearly 5%. Emerging markets gained nearly 9%, their strongest January in years.
Energy led all sectors with a gain of more than 14%. Financials were the weakest sector, down more than 2%.
These trends didn’t just continue into February. They intensified.
As of February 21, 2026, the S&P 500 is up just 0.9% for the year. But the S&P 500 Equal Weight Index, which gives the same importance to every stock in the index, not just the largest, is up 6.4%. That 5.5-percentage-point gap tells you something important: the mega-cap stocks that dominated returns for years are now dragging the index down, while everything underneath them is performing well.
Small-caps are up more than 9% year-to-date. Mid-caps are up more than 9%. Value is beating growth by nearly six points. Energy stocks are up 22%. Materials are up 16%. Industrials are up 14%. Gold and silver stocks are up 26%.
Meanwhile, information technology is down 3.5%. Consumer discretionary is down 3.4%. The Nasdaq is down 1.5%.
These are not small differences. They represent a fundamental reassessment by the market of which companies deserve premium valuations and which don’t.
The Split Inside Technology
If the broad rotation from growth to value is the headline, the story within the technology sector is the real revelation.
The S&P Information Technology sector is down 3.5% year-to-date. But that number conceals an extraordinary divergence between two types of technology companies: those that make the physical infrastructure of the AI economy, and those whose software businesses AI may be about to disrupt.
On the losing side, and these are staggering declines for less than two months of trading, Intuit is down 43%. Gartner is down 39%. AppLovin is down 38%. Workday is down 36%. ServiceNow is down 32%. Salesforce is down 30%. Adobe is down 26%. Oracle is down 24%. Palantir is down 24%. Microsoft is down 18%. IBM is down 13%.
These are not obscure companies. These are the pillars of enterprise software, the CRM platforms, the workflow automation tools, the consulting firms, the cybersecurity providers. The market is asking a simple and devastating question: if AI can automate workflows, write code, manage customer relationships, and analyze data, what happens to the $300-per-seat software license?
On the winning side: SanDisk is up 174%. Teradyne is up 68%. Western Digital is up 66%. Corning is up 59%. Micron Technology is up 50%. Seagate is up 49%. Applied Materials is up 46%. Lam Research is up 43%. These are the companies that make the chips, the memory, the semiconductor manufacturing equipment, the connectors, and the storage devices that AI physically runs on. The market is rewarding the companies selling picks and shovels into the AI gold rush.
And NVIDIA, the bellwether of the entire AI trade, the company whose GPUs power virtually every major AI model in production, is up 2% year-to-date. Flat. Right in the middle.
The market isn’t rejecting AI. It’s repricing who wins from AI.
Replace or Augment: That’s the Question
What makes this market rotation so consequential is that it’s happening against the backdrop of the most important economic debate of our time: what does artificial intelligence actually do to the economy, and who benefits?
At the skeptical end, Daron Acemoglu, the MIT economist who won the Nobel Prize in Economics in 2024, estimates that only about 5% of jobs will be meaningfully affected by AI over the next decade, and that the GDP boost will be closer to 1% than the double-digit numbers the optimists project. His central concern is what he calls “so-so automation,” technology that replaces workers without actually improving productivity. Think self-checkout kiosks: they eliminate the cashier but don’t make the shopping experience faster or better. Acemoglu warns that without deliberate policy choices, the gains from AI will accrue to capital owners while workers lose income and bargaining power.
At the other end, Goldman Sachs has estimated that widespread AI adoption could raise U.S. labor productivity by 15% over the coming decade. The World Economic Forum projects 170 million new roles created globally by 2030 against 92 million displaced, a net positive of 78 million jobs, but with massive workforce disruption in between.
The most nuanced thinking on the economics sits with Erik Brynjolfsson at Stanford’s Digital Economy Lab, probably the leading academic economist working on technology and productivity. Brynjolfsson sees the promise but warns about what he calls the “Turing Trap.” His argument is that when AI is designed to mimic and replace human workers, it concentrates wealth and power among those who control the technology. When it’s designed to augment human capabilities, making a doctor better at diagnosis, making an engineer more productive at solving problems, making a financial advisor better at analysis, it creates far more value and distributes it more broadly.
This distinction between replacement and augmentation isn’t academic. It’s the dividing line in the market data we just looked at. ServiceNow down 32% is the market pricing in replacement, the possibility that AI doesn’t need a workflow automation platform because AI is the workflow automation. Micron up 50% is the market pricing in augmentation, more humans using more AI tools that need more memory and more compute.
Brynjolfsson’s recent work with the Stanford Digital Economy Lab reinforces this. His research shows AI tools already lifting lower-skilled workers to perform at higher skill levels, which he views as genuinely constructive. A customer service representative with an AI copilot handles complex cases that previously required a specialist. An analyst with AI tools produces research that previously took a team. The capability floor rises. That’s augmentation at work, and it’s happening now.
But Brynjolfsson also talks about a “productivity J-curve” that every investor should understand. Firms investing heavily in AI see a short-term productivity dip before the gains materialize, because the organizational restructuring required to fully leverage AI takes time. New technology demands new processes, new training, new ways of thinking about workflows. Most companies are still on the downward slope of that J-curve, spending heavily, reorganizing painfully, and waiting for the payoff. This is exactly what the market is expressing when it rotates away from companies whose valuations depend on AI productivity gains that haven’t arrived yet and toward companies with earnings that exist today.
And then there is Dario Amodei, the CEO of Anthropic, who occupies a category of his own. In October 2024 he published a 15,000-word essay called “Machines of Loving Grace” that laid out what the world could look like if AI goes right: a century of medical and scientific progress compressed into a decade, infectious diseases largely conquered, economic growth accelerated across the developing world. But the essay was framed not as a prediction but as a conditional, what happens “if everything goes right.” And Amodei himself has said that most people are underestimating just how radical the upside of AI could be, just as they’re underestimating how bad the risks could be.
That duality defines him. He built one of the most powerful AI systems in the world, and then spent the next year fighting to make sure someone is paying attention to what happens when it’s deployed without guardrails. He warned in January 2026 that “humanity is about to be handed almost unimaginable power” and questioned whether our institutions have the maturity to wield it. He compared selling advanced AI chips to China to selling nuclear weapons to North Korea. He told 60 Minutes he is “deeply uncomfortable” with these decisions being made by a few companies and a few people.
Amodei is not the optimist and not the skeptic. He is the builder who sees both the ceiling and the floor, and that perspective puts him at the center of the most important collision in this story.
The Regulatory Vacuum
If the economic question is “what will AI do?”, the political question is equally urgent: who decides how it’s deployed?
The Trump administration has taken a clear position: get out of the way. Executive orders in 2025 revoked the Biden-era AI safety framework, directed agencies to remove regulatory barriers, and in December established a Department of Justice AI Litigation Task Force to challenge state-level AI laws. The administration is using $42 billion in federal broadband funding as leverage, states with AI regulations deemed “onerous” risk losing infrastructure grants. David Sacks, the administration’s AI and crypto czar, has singled out Colorado’s algorithmic discrimination statute as “probably the most excessive” and accused Anthropic of running a regulatory strategy based on “fear-mongering.”
And the collision is now tangible. As of this week, Anthropic’s $200 million Department of Defense contract is under review after the company raised concerns about how its AI technology was being used in military operations. Anthropic’s contract prohibits use in lethal military applications and bans mass surveillance of Americans. Among all the major AI companies contracting with the government, only Elon Musk’s xAI has given the military unrestricted use of its models.
Meanwhile, Europe, which passed the most comprehensive AI regulation in the world with its EU AI Act, announced in November that it would delay enforcement of its high-risk AI rules by sixteen months. The standardization bodies tasked with developing technical compliance standards missed their deadlines. Europe’s own infrastructure for implementing the rules it wrote isn’t ready.
So the United States is deregulating. Europe is stalling. And the most consequential question about the most powerful technology since the internet, who profits from it, who gets displaced by it, and what guardrails apply, is being decided not by Congress, not by voters, but by a handful of executives and one White House advisor. Whether that’s the right arrangement depends on how much you trust the people in those chairs.
What It Means for Investors
The convergence of these threads, the market rotation, the AI economic debate, and the regulatory vacuum, leads to a few observations that matter for portfolio construction.
First, concentration risk is real and it’s showing up now. For years, a portfolio heavily weighted toward U.S. large-cap technology was the right call. It may still be over the long term. But the last two months demonstrate how quickly the narrative can shift. A portfolio that includes international stocks, small-cap stocks, value-oriented holdings, and real assets like energy and materials has significantly outperformed a U.S. large-cap growth concentration in 2026.
Second, the market is distinguishing between AI hype and AI reality. Companies with demonstrated earnings and cash flows are being rewarded. Companies whose valuations depend on AI delivering projected productivity gains are being scrutinized. This is healthy. It doesn’t mean AI won’t transform the economy, it may well do so. But the market is demanding evidence.
Third, fixed income continues to do its job quietly. The Bloomberg U.S. Aggregate Bond Index returned 0.11% in January, but municipal bonds returned nearly 1%, and TIPS outperformed nominal Treasuries as inflation expectations edged higher. With the 10-year Treasury at 4.24% and the 2-year at 3.54%, the yield curve is offering reasonable compensation for fixed-income investors willing to be selective about credit quality and duration.
Fourth, diversification, the kind that felt painful during the years when U.S. large-cap technology was the only thing working, is paying off. Emerging markets are up nearly 9% for January alone. International small-cap value delivered more than 7%. The gap between global diversification and U.S. concentration hasn’t been this wide in years, and it’s running in favor of diversification.
We don’t know whether AI will deliver the transformative growth the optimists promise, the modest disruption the skeptics expect, or something in between. We don’t know whether the regulatory approach will be hands-off or prescriptive or some combination that hasn’t been articulated yet. What we do know is that the market is actively repricing these uncertainties, not in theory, but in the daily movement of real money.
The best response to genuine uncertainty isn’t prediction. It’s preparation. A broadly diversified portfolio, with appropriate exposure across sizes, styles, geographies, and asset classes, doesn’t require you to be right about which scenario unfolds. It positions you to participate in whichever one does.
Mary Shelley figured this out in 1818. Victor Frankenstein’s problem was never building the thing. It was everything that came after, the part he never planned for. Two centuries later, we’re watching the same story play out in real time. The creation is extraordinary. The question is whether the people who built it, and the people who regulate it, and the people who invest around it have a framework for what comes next. That’s not a technology question. It’s a human one.
Regards,
John Gorlow
President
Cardiff Park Advisors
888.332.2238 Toll Free
760.635.7526 Direct
760.271.6311 Cell
Past performance is no guarantee of future results. Index returns are for illustrative purposes and do not reflect actual fund performance. You cannot invest directly in an index. The opinions expressed are those of Cardiff Park Advisory and are subject to change without notice. This material is for informational purposes only and should not be considered investment advice.
Sources: Barchart, FactSet, Bloomberg, Goldman Sachs, IMF, World Economic Forum, Stanford Digital Economy Lab. Market data as of February 21, 2026.