US tech stocks are caught in an unusual tug-of-war right now: AI fundamentals keep strengthening, yet macro pressures from a firmer yen, rising interest rates and wider credit spreads are all building — while the Nasdaq 100 stubbornly refuses to break down.

For tech equities, the real danger isn't fading AI demand. The actual threat lies in the possibility that macro conditions and systemic leverage suddenly shift, triggering a rapid market repricing.

Goldman Sachs trading desk head Rich Privorotsky notes that frontier AI models continue to push capability boundaries. Google Cloud and AI server demand remain robust, and Broadcom even suggests actual demand has already exceeded its US$115 billion fiscal 2027 projection. Meanwhile, the yen's appreciation, widening tech CDS spreads and rising Treasury yields all point to tightening financial conditions.

When fundamentals march higher, macro signals deteriorate, and price refuses to confirm either side, markets are often storing energy for a sharper directional move. The volatility market echoes the same message: six-month realized correlation on the S&P 500 has fallen to 11, and SPX straddle prices sit near year-to-date lows, implying the market is pricing in only modest expectations of a large swing ahead.

AI bull thesis intact — the macro side is the real risk

AI fundamentals continue to strengthen. Privorotsky points out that frontier models are now solving math and science problems previously deemed intractable, including OpenAI's proposed solution to the Navier-Stokes equations. The new Astra model has been trained on more than 100,000 GPUs, with peak power consumption of 70–100 megawatts, and its capabilities clearly surpass previous generations.

Demand shows no meaningful cooling either. Google Cloud's single AI orders above US$100 million have more than doubled both quarter-over-quarter and year-over-year. Payback periods for AI server investments have compressed to under two years — and shrink to roughly one year with custom silicon.

Broadcom maintains that real AI demand already exceeds its fiscal 2027 outlook of US$115 billion. The factors now constraining industry growth are shifting from insufficient demand to power availability, data center location and supply chain capacity.

So if tech stocks turn lower from current levels, the most likely culprit isn't a sudden collapse in AI demand — it's interest rates, the yen and leverage all pressing at once.

Yen strength turns carry trades into a hidden risk

The yen's appreciation is becoming one of the market's most critical macro variables.

With traders increasingly pricing in further Bank of Japan tightening and Japanese capital repatriation, the yen keeps climbing. US Treasury Secretary Bessent recently struck a hawkish tone, saying, "I'm the house now... you can bet against me all you want." But that has done little to reverse the yen's upward trajectory.

What deserves close attention is yen-funded carry trade unwinding. If Japanese capital accelerates its return home, Japanese bonds and equities could absorb more funds while simultaneously forcing deleveraging in US markets.

Historically, yen strength correlates with falling risk appetite and rising volatility. But this time, the Nasdaq 100 has shown greater resilience to yen appreciation. Whether that means tech stocks are genuinely strong, or macro pressures haven't fully transmitted yet, remains to be seen.

CDS spreads widen — stocks refuse to confirm

The credit market is already flashing a clearer warning.

Since mid-August, tech CDS spreads have steadily widened. When CDS spreads blew out rapidly in July, the Nasdaq 100 saw a synchronous pullback. But this time around, credit spreads are widening again while equities show barely any response.

The credit market is signaling caution — the stock market is declining to confirm.

The rates market shows a similar pattern. Treasury yields keep climbing, which should further pressure richly valued tech stocks, yet mega-cap tech has held its ground. In the short term, that reflects strong market absorption capacity. But if yields continue higher, the pressure currently absorbed by prices could be released in a concentrated burst.

Extreme volatility compression hints at an explosive breakout

More than direction, the key question right now is: just how compressed has volatility become?

Six-month realized correlation on the S&P 500 has dropped to 11. Over the past 25 years, similar levels appeared only in February 2007 and January 2018 — the latter preceded the "Volmageddon" episode. Analyst Garrett notes that such extreme low-correlation states rarely persist for long, and historically tend to end with a fast reversion in index volatility and convexity.

At the same time, SPX straddle pricing remains near yearly lows, suggesting the market is assigning a low probability to any large swing ahead.

The current setup is a rare combination: AI fundamentals strengthening, macro and credit markets flashing warnings, the Nasdaq 100 refusing to pick a direction, and volatility pinned at depressed levels. Once price finally breaks out, the move could come with a violent spike in volatility, producing a sharp one-way trend.