Citrini Declares 'End of the AI Trade as We Know It' — Asia's Markets Agree

Asian markets moved exactly as Citrini Research's new report predicted: Korea's Kospi plunged 5% on Thursday while Hong Kong's Hang Seng rallied 3.6%, as traders dumped AI chipmakers and bought consumer and internet stocks.

The rotation hit the sector's bellwethers hard. Memory-chip makers Samsung Electronics and SK Hynix tumbled, while Tencent and Alibaba both spiked more than 6%. In U.S. premarket trade, Intel and Iren fell 2% and Nvidia lost 1%.

The driver was "State of the Themes," published Wednesday by Citrini Research, an independent boutique whose thematic reports have made a big splash in 2026 — including a field trip to Hormuz and a futuristic look at what AI might mean in practice. The report's timing was sharp: it dropped in the immediate aftermath of the midterm elections and made one bold prediction, "peak capex growth."

Why does Citrini think the AI trade is over?

The argument reduces to two numbers. Annualized hyperscaler investment is approaching $1 trillion, while annual recurring revenue across the entire AI ecosystem sits around one-fifth of that. "Clearly something has to give," Citrini observes — and it probably involves both cutting the capex and lifting the revenues.

The firm's verdict on the trade itself is blunt: whatever anyone thinks about AI, it's not the table-thumping buy it was.

Investors face a second problem. Many of the AI bets they're being forced to make are becoming longer-term, gambling on the total addressable market of 2030. Meanwhile, hyperscalers are sucking available capital out of the market through huge equity and bond offerings.

What changed with the midterms?

After the Democrats' commanding victory, Citrini sees additional physical risks to executing the AI trade. Data-center construction will, at the very least, become more expensive and complicated, the report said. That puts more question marks beside the bottlenecks trade in power and chips.

Citrini is also deterred by the token price wars between OpenAI and Anthropic that began this summer. The wars confirmed the firm's fears of commoditization and of consumer price elasticity — how sensitive buyers are to changes in cost. Its analysis finds AI is now starting to present headwinds through competition and disintermediation that undermine the software-as-a-service sector.

How should investors reposition?

Citrini's track record gives its calls weight. The firm's flagship Citrindex model portfolio has generated a total return of 257% since May 31, 2023 — more than 2.5 times that of the S&P 500.

The team thinks it's time to look for relative value rather than outright direction. The portfolio is being adapted toward reversals, pair trades, lower-beta plays and overlooked real-economy sectors: "Boomer software," financials, entertainment and energy.

The suggested pair trades include:

• Long Nvidia against short Advanced Micro Devices
• Oracle paired against Nebius
• Samsung against SK Hynix
• Meta versus Alphabet

Given the relative strength of the economy, better wages and sticky inflation, Citrini favors exploiting that environment with consumer names. It points to Victoria's Secret and e.l.f. Cosmetics. Entertainment exposure comes through media stocks like IMAX and Cinemark, plus live events via MGM Resorts.

The market's verdict on Thursday suggests investors didn't need much convincing. If Citrini's capex math is right, the question for the rest of 2026 is whether hyperscalers cut spending first — or keep raising capital until the revenue gap closes the hard way.