← All posts / Industry

Two Blogs Became One Research Empire: SemiAnalysis Buys Citrini, the Substack That Once Melted Software Stocks

Dylan Patel's SemiAnalysis has acquired Citrini Research, the 260,000-subscriber Substack whose '2028 Global Intelligence Crisis' memo triggered a Wall Street selloff — a first consolidation in independent AI research.

Two Blogs Became One Research Empire: SemiAnalysis Buys Citrini, the Substack That Once Melted Software Stocks

In February 2026, a Substack post did something that earnings misses and Fed speeches had failed to do: it knocked the wind out of Wall Street. “The 2028 Global Intelligence Crisis,” published by a then-obscure research shop called Citrini Research, sketched a scenario in which AI-driven white-collar displacement overwhelmed the economy by mid-2028. Software stocks — including names like DoorDash and BlackRock-adjacent platforms caught in the “AI loser” bucket — sold off hard within days. Bloomberg reported at the time that founder James van Geelen was genuinely shocked his hypothetical had spurred a market meltdown.

Seven months later, the punchline landed: Bloomberg confirmed on Friday, September 11, that Citrini Research has been sold to SemiAnalysis, the semiconductor and AI-infrastructure research powerhouse founded by Dylan Patel. Financial terms were not disclosed. Van Geelen will remain as Citrini’s CEO for now, with the brand keeping a degree of operational independence, and — according to multiple reports — he is planning to launch a new fund. Two of the most closely read independent research shops in the AI era are now under one roof.

What each side brings to the table

On the surface, this looks like a niche media acquisition. It is anything but. Both firms sit at the exact center of the most important question in public markets right now: how does a multi-trillion-dollar AI infrastructure build-out actually flow through to earnings, margins, and stock prices?

SemiAnalysis, founded by Dylan Patel in 2020, grew out of hardware forums and blogs into an institution with over 50 employees (some reports put headcount near 90) and more than 180,000 subscribers. Its reports on GPU supply chains, custom silicon roadmaps, and data-center economics have become required reading for chip executives and institutional investors. The firm has a track record of consequential calls — in 2020 it publicly predicted that MediaTek, not Qualcomm, would inherit the TSMC volume Huawei lost to sanctions, a call that aged remarkably well. By 2026, NVIDIA’s GTC keynote was name-checking Patel and showcasing the firm’s performance rankings on stage. SemiAnalysis has also been transparent about its own role as a demand-side data point: internal AI spending has reportedly risen from under $100,000 a year ago to roughly $11 million.

Citrini Research is the smaller, nimbler, and arguably more controversial half. Founded in New York in 2023, it grew to approximately 260,000 Substack subscribers on the strength of thematic long/short analysis that tracks how AI infrastructure spending translates into public-equity moves. Van Geelen’s path was unconventional — from biology and psychology at UCLA into finance and healthcare themes — and it shows in Citrini’s willingness to publish narrative-driven scenario work that traditional sell-side research would never touch. The firm had raised roughly $5.05 million in cumulative financing through December 2025, according to SEC filings, tiny money by AI-era standards.

Why this deal makes sense

The strategic logic is straightforward: the two firms cover opposite ends of the same trade. SemiAnalysis tells you whether the wafer allocation, the HBM supply, and the gigawatt build-out are real — the supply-side truth of the AI boom. Citrini tells you how the market will react when the spending ripple reaches software margins and labor markets — the demand-side narrative. A buy-side institution that wants both the build list and the stock map can now buy them on the same invoice.

That last point is the quiet revolution here. Independent research — once dismissed as subscription newsletters run out of bedrooms — has become an asset class that can be acquired, rolled up, and scaled. When a single Substack post can shave billions off software market caps, the research itself carries balance-sheet weight. The acquisition signals to every independent analyst with a audience that their shop is now a potential target, and to every hedge fund that the pitch “we read both the chip supply chain and the software P&L” is now a productized offering.

The open questions

The undisclosed price leaves real gaps. How deeply will Citrini be integrated — does “operational independence” survive the first quarter? More pointedly, van Geelen’s plan to launch a new fund raises a classic conflict-of-interest question: can a research shop that moves markets also run money off its own calls, under the same corporate parent as a supply-chain research brand? The SEC has no dedicated framework for “Substack-driven market influence,” and the combination of research, subscriptions, and fund performance splits — none of which have been disclosed — will be watched closely.

There is also a competitive read. Investment banks spending millions on sell-side chip research now compete for attention with a merged entity that combines SemiAnalysis’s component-level teardowns with Citrini’s thematic positioning power. If the pattern holds, this is the first platform-level merger in independent AI research — and unlikely to be the last. The market is centralizing pricing power over the AI narrative into fewer hands.

For now, one thing is certain: the next research note that moves software stocks will carry footnotes from the chip supply chain. The author lists just merged.

The bigger picture

This deal is a marker of how quickly the AI information economy has matured. In 2020, SemiAnalysis was a blog; in 2023, Citrini was a solo Substack. In 2026, one acquired the other, with a combined reach approaching half a million subscribers and influence that reaches into NVIDIA keynotes, hedge fund positioning, and — at least once — the broad software sector’s market cap. Research, it turns out, is one of the highest-leverage products of the AI boom: it costs almost nothing to produce relative to the capital it moves.

Whether van Geelen’s new fund becomes the template (analyst-turned-PM, like so many before him) or he eventually exits to build something new, the Citrini-SemiAnalysis combination has already changed the shape of independent research. The era of the solo newsletter tycoon is giving way to the era of the research holding company.