Nvidia insider sells as "Fab 10" reshapes $3 trillion AI chip trade

A notable Nvidia insider sold their stake before the reason emerged: the "Fab 10" — a new group of AI infrastructure players — is reshaping the $3 trillion chip investment trade, signaling capital rotation away from single-stock dominance.

The sale, executed quietly in recent weeks and disclosed on June 18, was tied to the emergence of a cohort of 10 companies that investors believe will capture a growing share of AI infrastructure spending as the market shifts from single-vendor dominance to a multi-player ecosystem. The insider, whose identity has not been publicly confirmed, sold a significant portion of their Nvidia holdings before the "Fab 10" thesis became widely discussed among institutional investors.

"The AI chip trade is no longer a one-stock story," said Rachel Kim, semiconductor supply chain analyst at Edgen. "The Fab 10 represents a structural shift where capital is rotating from a single dominant player into a basket of infrastructure beneficiaries — foundries, networking, memory, and custom silicon designers."

The rotation comes as JPMorgan projects more than $3 trillion in AI chip and hardware financing over the next five years, with silicon spending rising to about $800 billion annually by 2030 from $340 billion in 2026. Nvidia remains the dominant force — it reported $81.6 billion in revenue for the fiscal first quarter, up 85 percent from a year earlier, and Chief Financial Officer Colette Kress has said AI spending is on track to reach $3 trillion to $4 trillion annually by the end of the decade. But the composition of that spending is shifting. JPMorgan strategist Tarek Hamid wrote in a June note that GPU and AI-specific chip spending could grow to 60 percent of total annual data center expenditure by 2030 from roughly 50 percent, with replacement demand becoming a key driver.

The "Fab 10" thesis argues that as AI workloads mature and hyperscalers like Amazon, Microsoft, and Google push in-house chip development, the addressable market broadens beyond Nvidia's GPU monopoly. Amazon is in talks to sell its own AI chips, according to reports, while Google's TPU shipments are expected to reach 8 million units next year, nearly matching Nvidia's projected 9.9 million GPUs. Nvidia still commands the highest-volume pipeline — 8.9 million GPUs expected to ship this year versus 4.5 million comparable TPUs from Google and 1.9 million Trainium and Inferenta chips from Amazon — but the competitive gap is narrowing.

For investors, the question is whether Nvidia's valuation already reflects this broadening landscape. The stock has gained more than 12 percent year to date, trailing CPU maker AMD, which has more than doubled. With Nvidia's data center revenue still growing at an 85 percent clip, the bull case rests on the company maintaining its roughly 80 percent GPU market share as the total addressable market expands. The bear case, now bolstered by insider selling and the Fab 10 narrative, is that capital will increasingly flow to a diversified set of AI chip winners — foundries like TSMC, networking specialists, memory makers, and custom silicon designers — compressing Nvidia's premium multiple over time.

This article is for informational purposes only and does not constitute investment advice.