Epoch AI: Frontier Labs Pivot to Sales as Research Share Shrinks
Summary
- • OpenAI and Anthropic now dedicate 28–31% of open roles to sales, up from 17–18% a year ago
- • Research share dropped to just 7–12% of open positions at both companies
- • Adoption-support roles doubled as enterprises struggle to deploy AI products effectively
- • OpenAI pursues 21 custom silicon roles; Anthropic has zero chip development hiring
Details
GTM now top hiring category at both Anthropic and OpenAI
Anthropic's GTM share: 17% → 31%. OpenAI's: 18% → 28%. Both companies now spend more hiring energy on selling and deploying AI than on building it. Research fell to 12% (Anthropic) and 7% (OpenAI).
Adoption roles doubled — deployment is the real bottleneck
Roles dedicated to helping customers use AI grew from 5% to 11% at Anthropic and 11% to 17% at OpenAI. Titles include AI Success Engineers, Solutions Architects, and Forward Deployed Engineers — indicating enterprise deployment difficulty is a growing constraint on revenue.
OpenAI betting on custom silicon; Anthropic staying software-only
OpenAI's 21 custom silicon open roles signal a push for chip independence, potentially reducing Nvidia dependency and lowering inference costs. Anthropic has no equivalent effort, reinforcing its identity as a safety-first software lab.
US market dominant; hardware ambitions diverge by lab
52% of Anthropic sales roles and 55% of OpenAI sales roles are US-based, suggesting North America remains the primary revenue focus. OpenAI and Google DeepMind are also investing in hardware (robotics, consumer devices); Anthropic is not.
Frontier Lab Strategy Analysis via Job Postings
What This Means
The data suggests the AI industry's near-term competitive battles will be won on go-to-market execution and customer success, not just model breakthroughs. For enterprise buyers, this means better deployment support is coming. For AI practitioners, it signals the industry is maturing from invention into commercialization — and customer-facing technical roles are now the fastest-growing career path at the frontier labs.
Sentiment
Broadly supportive of the pivot to commercialization and adoption, seen as industry maturation
“Adoption and diffusion are bigger barriers to AI transformation right now than raw capability improvement... the gap between what AI can already do and what people are using it for is already gigantic.”
“This is an ingenious way to predict the roadmap of the AI Labs.”
“Yeah definitely one the most informative source.”
“the shift from research to GTM roles tells the real story: the model moat is shrinking, the distribution moat is everything.”
Split
~90/10 supportive of commercialization focus vs. concern over research slowdown.