UK-based Graphcore, maker of machine learning chips, has raised $50M from Sequoia Capital
UK chipmaker Graphcore has raised $50m from Sequoia Capital, one of Silicon Valley's top venture capital firms, as the start-up angles to become Britain's biggest maker of artificial intelligence hardware.
Context & Ripple Effects
Sequoia's $50M check in November 2017 was the founding bet on Britain's answer to Nvidia: a Bristol startup building the IPU, a processor designed specifically for machine learning rather than repurposed graphics silicon. The round put Graphcore on the path that led through a $200M raise at a $1.5B valuation just over a year later, with Microsoft and BMW joining the cap table.
Read against the full arc, this is the entry point of one of Europe's most-funded AI chip stories — and its cautionary coda. By 2023 Graphcore had raised roughly $730M while reporting only ~$5M in 2021 revenue and a $183M loss, with investor Microsoft reportedly declining to use its chips; by early 2024 it was exploring a sale to foreign owners at a price that could exceed $500M but sat far below its $2.77B peak.
First-order effects
- Sequoia gains an early-stage position in AI accelerator silicon at pre-scale pricing, before the category's valuations inflated — the same firm later managing $56B in assets and netting ~$3B from the reported Wiz sale to Google.
- Graphcore converts Silicon Valley validation into hiring and IPU development capital, letting it attack Nvidia's training market from the UK rather than competing for US talent alone.
Second-order effects
- The Sequoia endorsement pulls in strategic money at escalating prices: Atomico, Sofina, BMW and Microsoft follow into the $1.5B round, then pension-plan capital leads a $222M Series E at $2.77B — each round raising the bar for commercial proof.
- Microsoft's dual role as investor and prospective customer becomes the pivotal test; when it reportedly declines to deploy Graphcore's chips, the startup loses the anchor-tenant revenue model that justified the valuation ladder.
Third-order effects
- If the pattern holds, specialist AI-chip challengers face a structural squeeze: hundreds of millions in venture capital cannot substitute for hyperscaler design wins, and exits clear well below peak marks — Graphcore's explored $500M+ sale versus its $2.77B high is the template.
- Britain's ambition to host a homegrown AI hardware champion ends up dependent on foreign acquirers — rumored buyers include OpenAI, SoftBank and Arm — making national tech-sovereignty claims contingent on deal outcomes rather than domestic scale.
The trend: Venture capital poured record sums into AI-silicon challengers during the last cycle, but funding scale outran commercial traction, leaving most merchant-silicon startups to exit below their peak valuations.