OpenAI says Samsung is helping build its next-gen chips
Two strands of the AI buildout surfaced this afternoon: OpenAI describing how far its silicon work with Samsung has actually gone, and a fresh challenge to the jobs story that countries are using to justify data centers.
OpenAI says joint chip production and research is now the deepest part of its relationship with Samsung. Harrison Kim, General Manager of OpenAI Korea, told a press conference in Seoul on Wednesday that "one of the areas where we have made the most progress and gained the most recognition with Samsung Electronics is our joint production and research on the next-generation chips we are developing" — without elaborating on what Samsung is making or when it ships. Samsung Electronics said it could not confirm information related to customers. Kim separately called Samsung "one of the largest-scale deployments of ChatGPT globally."
The context matters more than the quote. OpenAI unveiled its first custom inference chip, Jalapeño, in June, designed with Broadcom and slated for manufacture at TSMC — we covered the early numbers in Jalapeño's first benchmarks put OpenAI on Nvidia's turf. Bringing Samsung into the next generation adds a second heavyweight on the manufacturing side and deepens a relationship that so far ran through memory: Samsung and SK Hynix signed letters of intent last year to supply memory for OpenAI's Stargate project. Our read: this is supply-chain hedging, not a Broadcom replacement story, and the vagueness is deliberate — no lab names a second source before it has to.
A security startup that maps AI agents alongside employees raised $25 million from Sequoia. Cymphony's Series A was co-led by Sequoia and SMBC Fin Atlas Beyond Fund, valuing the New York- and Tel Aviv-based company at more than $100 million, on roughly $30 million raised to date. Its pitch is a "workforce graph" that gives security teams one view of employees, agents, and other non-human identities and the data each can reach — because agents rarely pass through the same access controls a human hire does.
The findings it cites are the reason the round clears our bar. At one U.S. public company, Cymphony says it found about 85,000 files that had become accessible to AI tools and agents, and verified none had been read through those systems; in another case an external collaborator installed an unsanctioned instance of Claude that used their existing access to scan thousands of sensitive files. CEO Shy Dekel's framing is the industry's in one line: "enterprise security was designed for human employees." The company reports seven figures in annual recurring revenue in its first year of sales, with KKR and Syngenta among customers — and it is racing Microsoft, Okta, CyberArk, Wiz, and Varonis to define the category.
UK data centers are on track to create roughly a quarter of the jobs the industry promised. Think tank Verdant puts current direct employment at UK data centers at about 4,400 roles against trade body techUK's 24,300, and projects around 10,400 permanent operational jobs by 2035 against techUK's 40,200. The gap comes down to a single multiplier: techUK assumes 8.5 jobs per megawatt, while Verdant, after examining 20 planned UK sites using planning applications and public staffing figures, calculates closer to 1.2 — newer, larger, more automated facilities need fewer permanent staff. techUK defended its methodology and called Verdant's sample too small; a UK government spokesperson disputed the report outright.
What to watch: whether the UK keeps selling data centers on jobs at all, or switches to the argument its own spokesperson reached for — compute, national security, and sovereignty.
If a data center creates one job per megawatt, should governments still be fast-tracking them on employment grounds? Tell us in the comments.
