The Tech and AI Rally That Refuses to Die: Why Record Yields Aren't Stopping the Party

By any traditional measure, tech stocks should be struggling.

The 10-year Treasury yield sits just below 5%, a level that historically acts as a gravitational pull on high-growth equities . The 20-year yield is even higher, hovering around 5.39% . The Federal Reserve just raised rates again in September 2026, and Fed officials are signaling more may come .

Yet the Nasdaq 100 is up more than 20% year-to-date, sitting within 1% of its record high . Semiconductor stocks have been on a five-session winning streak. AMD just crossed the $1 trillion market cap threshold . Micron joined the trillion-dollar club earlier this year after a stunning 730% rally in twelve months .

This is not supposed to happen. But it is happening. And the reasons why tell us something important about how markets actually work in this new era.

The Old Rulebook

The textbook logic is straightforward: when interest rates rise, the present value of future cash flows falls. Tech companies, whose valuations depend heavily on profits expected years down the road, are supposed to be hit hardest. Higher rates mean higher discount rates, which mean lower valuations for growth stocks .

This is why, for most of 2025 and early 2026, every uptick in Treasury yields sent tech stocks tumbling. Investors treated the 10-year yield as a ceiling—a level above which tech simply couldn't breathe.

That relationship has broken down.

What Changed

First, earnings caught up to the narrative.

The AI trade is no longer just a story about what might happen someday. It is now a story about what is happening right now. FactSet estimates that S&P 500 earnings are on track to rise nearly 29% in the third quarter, marking the third consecutive quarter of annual earnings growth exceeding 25% . Companies mentioning "AI" on earnings calls have seen higher average price increases than those that haven't .

Bloomberg's analysis found that the AI rally of 2025 was "underpinned by fundamental earnings growth rather than valuation expansion alone" . This is not the dot-com bubble. These companies are generating real cash.

Nvidia is forecast to generate $168 billion in revenue in 2026. The six largest US hyperscalers are projected to spend over $575 billion on capital expenditure this year . TSMC, the world's most important chipmaker, announced a record $52-56 billion capital spending plan for 2026—and its CEO openly acknowledged the risk: "If we don't do it carefully, that'd be a big disaster for TSMC" .

That kind of spending is not speculative froth. It is infrastructure investment at a scale that reshapes entire economies.

Second, the rate hike was already priced in.

When the Fed raised rates in September, the market didn't panic. It rallied. The reason: investors had already adjusted their portfolios for the hike before it happened . Inflation data, oil prices, and hawkish Fed rhetoric had been building for weeks. By the time the decision arrived, the bad news was in the price.

This is a crucial insight. Markets don't react to events; they react to events relative to expectations. The rate hike was not a surprise. What mattered was what came next—and what came next was Meta's new AI agent, Muse, which hit the top of Apple's App Store and sent Meta's stock up 11% in a single day .

Third, the AI boom is creating its own demand cycle.

Something unusual is happening in the semiconductor industry. Companies are raising prices because they can't keep up with demand. DRAM revenue grew nearly 60% quarter-over-quarter, driven primarily by price increases rather than volume . High-bandwidth memory, the specialized chips needed for AI training, is in such short supply that manufacturers have pricing power they haven't enjoyed in years .

This is the opposite of a demand problem. It is a supply constraint—and supply constraints create pricing power, which creates earnings, which justifies valuations even at elevated discount rates.

The Great Divergence

But here is where the story gets more complicated. The AI rally is not a rising tide lifting all boats. It is a sorting mechanism, and it is becoming more selective by the month.

The market is rewarding infrastructure—the picks and shovels of the AI gold rush—while punishing applications that haven't yet proven they can monetize the technology.

US semiconductor companies are expected to grow earnings by 98% in 2026, more than five times faster than the hyperscalers that are their biggest customers . Memory chipmakers, chip equipment manufacturers, and data center builders are seeing their earnings estimates revised upward at a pace that makes the broader market look stagnant.

Meanwhile, software and media companies have been hit hard. J.P. Morgan notes that software stocks are down 3% and media companies down 32% since the start of 2025, as investors worry about AI-driven disruption to their business models .

The market is making a bet: AI will create enormous value, but the value will accrue first to the companies building the physical infrastructure, and only later—if at all—to those trying to sell AI-powered services to consumers.

The Risks Nobody Is Talking About

The rally is impressive. It is also fragile in ways that are not immediately obvious.

The funding question. The AI buildout is being financed with debt. Hyperscalers have burned through their cash piles—AI capital expenditure has gone from 33% of their operating cash flow in 2023 to an estimated 93% in 2026 . When cash runs out, companies borrow. When rates are at 5%, borrowing is expensive. If rates go higher, the math gets harder.

The circularity problem. Nvidia sells chips to hyperscalers. Hyperscalers build data centers. Data centers need power. Meta is signing nuclear power deals for 6 gigawatts of electricity . This is a supply chain that consumes enormous capital before it produces a single dollar of AI revenue. If demand slows at any point in that chain, the whole structure wobbles.

The concentration risk. The AI trade has become the market. With SpaceX's IPO, the ten most valuable US public companies are now entirely tech companies . When one sector dominates this completely, the entire market becomes hostage to its fortunes.

What This Means for Investors

The rally is real. The earnings are real. The infrastructure buildout is real. But the easy money—the phase where you could buy anything with "AI" in the description and watch it double—is probably over.

What comes next is a stock picker's market, even within AI. The market is increasingly differentiating between companies with current earnings and companies with future promises . The former are holding up despite high rates. The latter are being sold.

For everyday investors, the lesson is not to chase the hottest AI stock. It is to recognize that we are watching a structural shift in the global economy—one where computing power is becoming as essential as electricity—and to position accordingly.

The rally may pause. It may correct. But the forces driving it—the buildout of AI infrastructure, the earnings growth of the companies enabling it, and the productivity gains it promises—are not going away because Treasury yields tick above 5%.

The old rules said tech can't thrive when rates are high. The new reality is more nuanced: tech can thrive when rates are high, as long as the earnings are real. The market is now sorting the real from the imagined. That is not a reason to fear the rally. It is a reason to understand it.