The Paradox Nobody Talks About

On September 16, 2026, something happened that finance textbooks tell us shouldn't happen. The Federal Reserve raised interest rates—its first hike since 2023, to a range of 3.75%–4.00%—an action that historically punishes technology markets. And yet, after the initial blow, the tech sector ended up leading: the Nasdaq closed up 1.7% and semiconductors up 3.1%. Growth stocks rose. And most analysts continued with their usual narratives, as if nothing extraordinary had just occurred.

But something extraordinary did occur. And it's not about the Fed. It's about what the Fed now represents: relative irrelevance in the face of a force nobody knows how to value.

The comfortable lie that ended

For decades, investors played the same game: low rates = tech grows. High rates = tech gets punished. It was simple. It was predictable. It was, fundamentally, incomplete.

What happened on September 16 is that the market finally acknowledged out loud what it already knew silently: tech valuation doesn't depend primarily on interest rates. It depends on a bet about who will control productivity in the next decade.

And that bet is called Artificial Intelligence.

When the Fed raises rates, it's trying to slow demand for money. It assumes this will pressure growth valuations. But that assumes "growth" is disconnected from fundamental changes in productive capacity. In 2026, that assumption is obsolete.

Big tech companies—Nvidia, Microsoft, Google, Meta—aren't being valued on financial utopias. They're being valued because they already have AI systems trained, deployed, and generating measurable returns. It's not 2015 speculation. It's 2026 productivity.

So: why do they go up when rates go up? Because the assets that need cheap credit to exist (pre-profitable startups, SPACs, speculative corporate debt) go up less. Tech rises in relative terms. And in a world where money is more expensive, what generates real cash flow has more value, not less.

Let's be honest about the mechanics of that day: much of the rebound came when U.S. Treasury yields and oil prices retreated after the decision—the very channel that links rates to valuations. A single day proves no thesis. What's revealing isn't the rebound, but its composition: with money more expensive, the market kept paying for real productive capacity—AI, memory, semiconductors—instead of punishing it across the board.

The condition that remains

The event of September 16 is over. The Fed's next move will come, generating fresh headlines. But the condition that event revealed—the disconnect between the price of money and the price of AI-dominated productive capacity—that remains. And it's a real problem for anyone with capital to deploy, because it implies three things simultaneously true.

First: tech valuations can keep rising even with high rates, because they're anchored in something different than traditional speculation.

Second: this doesn't mean tech is safe. It means its risk is now idiosyncratic. Not market risk. The risk that AI doesn't deliver, or concentrates in so few companies competition becomes irrelevant.

Third: valuation criteria that worked for 20 years—comparing multiples, discount rates, profitability ratios—become less useful when the ground beneath you is shifting.

What does this mean for you?

If you're a capital manager or have significant assets to deploy, here's the real problem: it's not enough to "have exposure to tech." You need to understand which part of tech has real AI returns, and which is still speculation.

The roughly US$3.5 billion that flowed into Bitcoin ETFs in September isn't coincidence. It's not just "crypto is in fashion again." It's investors saying: "If traditional metrics are strained, maybe I need assets that don't depend on cash flow valuation." That's not prudence. That's unease. And collective unease is information.

Three structural questions you should ask yourself. First: where does my expected tech return come from—expanding multiples, or real returns that AI generates? If you can't tell the difference, you're speculating. Second: how much concentration in "AI winners" can I tolerate if 60% of my tech portfolio is in three companies and one has a compliance, cybersecurity, or regulatory event? Third: what's my firewall if the cycle turns and an event triggers demand for cash—do I have real shelter, or just assets that went up when everything went up?

The event and the condition

The Fed raised rates on September 16. That was an event. It passed. It will be followed by others. But the condition it revealed—that we live in a world where productivity matters more than the price of money—that doesn't pass. It remains. And it will remain as long as AI stays the scarcest asset in the economy: scarcer than capital, scarcer than money supply, scarcer than human capacity to generate ideas.

When that scarcity resolves—when AI becomes utility, when anyone can train it, when value distributes widely—then the rules change. Until then, the question that matters isn't "will tech go up or down?" It's: "do I understand who captures real value in this transition, and do I have the nerve to hold it if I'm wrong on timing?"

If the answer is no, maybe you've already spent too much on the wrong answer.

— R. B.