VC
Value Add VC
⚡HomePulse⚡Helpful Apps📝Blog
← Value Add PulseBIG TECHEarly-stage, no design contract yet

Anthropic in Early Talks With Samsung on Custom 2nm AI Chip

Anthropic is in preliminary discussions with Samsung to manufacture a custom AI accelerator on Samsung's 2nm process, joining a wave of frontier labs building their own silicon to cut Nvidia dependence.

Samsung SF2 (2nm)
Target Process Node
~15% vs FinFET
Performance Gain Cited
Stalled early June
OpenAI-Samsung Talks
Microsoft, Fractile
Other Chip Partners
TC
Trace Cohen
Early-stage VC & angel · Founder, New York Venture Partners
July 2, 2026
2 min read
ShareXLinkedInEmail
THE RUNDOWN
1

The talks target Samsung's SF2 2nm node, which uses Gate-All-Around transistors for roughly 15% better performance or meaningful power savings versus the prior FinFET generation

2

Anthropic recently hired Clive Chan, an early member of OpenAI's own custom-chip team, signaling a deliberate silicon-engineering buildout rather than a one-off exploration

3

Samsung's talks with OpenAI on a similar ARM-based inference chip reportedly stalled in early June over strategic differences, making Anthropic a potential replacement anchor customer

4

Anthropic is simultaneously in discussions with Microsoft and UK chip startup Fractile, meaning Samsung is one option among several rather than an exclusive partner

TC
The VC Read · Trace's TakeTrace Cohen

Every frontier lab now wants to own a sliver of its own silicon roadmap, and Anthropic hedging across Samsung, Microsoft and a scrappy UK startup like Fractile at once tells you this is about optionality, not conviction in any one partner. The real signal is the Clive Chan hire -- that's not exploratory, that's a lab building an internal chip organization the way it built its research org. Watch OpenAI's stalled Samsung talks closely; if Anthropic's approach was opportunistic, it's a small but real tell about how competitive the AI-lab relationship with foundries has become.

Anthropic has opened early-stage discussions with Samsung Electronics about manufacturing a custom AI accelerator on Samsung's 2nm SF2 process, according to The Information, as the AI lab pushes further into designing its own silicon rather than relying solely on Nvidia GPUs. The project is nascent -- no detailed chip design or manufacturing work has begun, and Anthropic may not proceed with Samsung specifically -- but the talks reflect a broader strategic shift among frontier labs toward vertically integrated compute.

The technical target is notable: Samsung's SF2 node uses Gate-All-Around nanosheet transistors, a structural departure from the FinFET transistors that dominated the last decade, allowing the gate to fully surround the channel for tighter electrical control. That translates to roughly 15% better performance at equivalent power, or meaningful power savings at equivalent performance -- either of which matters enormously at the scale of a frontier AI lab's inference and training fleet.

Anthropic isn't approaching this cold: the company recently hired Clive Chan, an early member of OpenAI's custom silicon team, as part of what The Information describes as a deliberate engineering buildout rather than opportunistic hiring. That matters because Samsung was already deep in talks with OpenAI on a similar ARM-based inference chip before those discussions reportedly stalled in early June over strategic disagreements -- leaving an opening that Anthropic's approach conveniently fills, whether or not that was the intent.

The custom-silicon trend now spans nearly every frontier lab: Google has TPUs, Amazon has Trainium and Inferentia, Meta has MTIA, and OpenAI has been pursuing its own chip program with Broadcom alongside the now-stalled Samsung talks. Anthropic entering this race -- and doing so through multiple parallel tracks with Samsung, Microsoft and Fractile rather than a single exclusive bet -- suggests the lab is treating chip diversification the same way it treats model architecture: hedge broadly, commit late.

The numbers in context: Nvidia still captures the overwhelming majority of AI training and inference compute spend industry-wide, and even successful custom-silicon programs at Google and Amazon took years to reach meaningful internal workload share. A 2nm Anthropic chip, if it happens at all, is realistically a 2028-or-later production reality, not a near-term Nvidia alternative -- this is a multi-year hedge, not an immediate supply shift.

For infrastructure investors, the read-through is that the AI compute stack is fragmenting one layer further: the labs that can afford custom silicon programs (Anthropic, OpenAI, Google, Meta, Amazon) are pulling further ahead of labs that can't, widening the moat between frontier-tier and mid-tier AI companies on cost-per-token alone. For chip-adjacent startups, Anthropic's multi-vendor approach (Samsung, Microsoft, Fractile) is a reminder that being one of several parallel bets is the realistic outcome, not an exclusive anchor deal.

What to watch: whether Anthropic's talks with Samsung progress past the exploratory stage into an actual design contract, what becomes of the stalled OpenAI-Samsung inference chip, and whether Fractile -- the smallest, least-resourced party in this three-way hedge -- can compete for meaningful allocation against Microsoft and Samsung.

ShareXLinkedInEmail
More onAnthropic →Nvidia →

Originally reported by The Information. Analysis and editorial commentary by Value Add Pulse.

← Back to Pulse

THE WIRE in your inbox

Tech, startup & VC news with Trace's take. Free, no spam.

Read Next

BIG TECH$2.5B / 6,000 employees

Microsoft Commits $2.5B and 6,000 Staff to New AI Deployment Unit

Microsoft is investing $2.5 billion to launch Frontier Company, a 6,000-person unit that embeds engineers directly with enterprise customers to get AI deployments actually working, following similar moves from Amazon, OpenAI and Anthropic.

BIG TECH480,126 vehicles delivered

Tesla Deliveries Jump 25% to a Record 480,126 Vehicles in Q2 2026

Tesla delivered 480,126 vehicles in the second quarter of 2026, a company record and a 25% jump from a year earlier, beating even the most bullish Wall Street estimates of roughly 418,000-420,000 by more than 60,000 units, according to Ars Technica and.

BIG TECH~700 satellites by July

Amazon Now Has Enough Satellites in Orbit to Launch Its Starlink Competitor

Amazon's Project Leo satellite broadband service has crossed the deployment threshold needed for commercial launch, The Verge reported July 2, with roughly 700 production satellites expected in orbit by the end of July after CEO Andy Jassy confirmed a 'mid-2026' launch target. The FCC waived Amazon's original July 30 deadline to have half its 3,232-satellite constellation in orbit, but is penalizing spectrum priority on satellites launched after that date to keep pressure on the rollout.

@Trace_Cohen·t@nyvp.com