OpenAI and Synopsys team up to build a chip-design model
A frontier lab just moved inside the chip-design toolchain, voice AI priced itself at $22 billion, and Google picked a winner in the assistant-format wars.
OpenAI and Synopsys signed a multi-year deal to build GPT-Synopsys, a model that runs Synopsys' chip-design tools. OpenAI licenses the electronic design automation software, hosts the model on its own infrastructure, and gets engineers to delegate design objectives and review the output — while customer data stays out of training and encrypted at rest. The economics are the interesting part: OpenAI pays a training subscription, and the two companies share revenue based on how well the model improves a design, tying OpenAI's payout to measured engineering results instead of seat licenses. Synopsys lifted its fiscal 2027 guidance to 15 percent revenue growth — well above the roughly 11 percent analysts had modelled — and its shares rose as much as 7 percent on the news. The guardrail is equally telling: every output still has to clear traditional sign-off tools, because, as Synopsys CEO Sassine Ghazi put it, the model needs those checks to verify the physics. This is the clearest signal yet that labs want to sell work inside expert software, not just answers beside it. We covered Synopsys' own agent push earlier this month — Synopsys puts 30 agents inside chip design — and bills by usage.
ElevenLabs let employees cash out stock at a $22 billion valuation — double the $11 billion mark from February. The $300 million tender offer was co-led by Wellington and T. Rowe Price, the kind of institutions that buy private shares expecting to hold them through an IPO, and it follows a $100 million tender at $6.6 billion last September. The New York- and London-based voice startup now counts among Europe's most valuable companies, and the speed of the re-rating — two doublings in a year — says voice agents are being priced as the interface layer of the AI stack, not a feature of it.
Google is retiring Gemini Gems for Skills, the agent-ready prompt format it inherited from Anthropic's open standard. Skills launch globally in Gemini chat today, existing Gems convert automatically, and support ends in November 2026 for personal accounts, March 2027 for most Workspace customers and June 2027 for education — Opal, Google's mini-app experiment, winds down in November too. Google's pitch is that skills can be chained, shared and triggered automatically when the assistant spots a matching request, which is exactly why the format is converging across the industry: if assistants are going to act, their instructions need to be portable files rather than chat memory. OpenAI is sunsetting its own Custom GPTs as well, which leaves Anthropic's open spec as the de facto common layer.
What to watch: whether GPT-Synopsys' early customer tests produce published design gains — and how Broadcom's custom inference chip work fits alongside it.
Is an AI that operates the design tools, with humans signing off at the end, something you'd trust with your next tape-out? Tell us in the comments.
Sources: Synopsys · Reuters · The Decoder · TechCrunch · Economic Times · Google · 9to5Google · The Decoder
