Normal Computing

Advanced Research + Invention Agency (ARIA) published its 2026 Corporate Plan, and Normal Computing's thermodynamic computing work is featured as a project spotlight. Building on the successful tape-out of CN101 last year, we're designing the next generation of Normal's thermodynamic architecture and deploying our purpose-built hardware for AI inference and long-context agentic reasoning into modern data centers, where different types of chips work together on increasingly complex AI workloads. The plan also details the ecosystem ARIA is building around low-power compute, including a new £50M Scaling Inference Lab backed by a further £20M pledge from the Department for Science, Innovation and Technology as part of the UK AI Hardware Plan. Our work has been supported by this partnership, and we're continuing to build collaborations across the semiconductor ecosystem. From the report: "The goal is to significantly reduce the power consumption of AI, driving a shift in our current computing paradigm by orders of magnitude. This will have a worldwide impact, and open a fundamentally new path forward." Read the full version here: https://lnkd.in/gZd2ZqhY

Several months ago we published a Verilog simulator built by AI agents in 43 days. The project is still running, with more than 10 agents writing simultaneously to a single shared branch. There are no PRs and no separate worktrees. The agents constantly bump into each other, and velocity holds anyway, because the agents resolving the conflicts are more capable than the conflicts themselves. We treat coordination as a temperature parameter. The project runs hot early, while the architecture is in flux and inconsistency is cheap, then anneals as the system matures and bugs take hours to trace. We call this thermodynamic programming, and the same structure appears in neural network training and collaborative writing. Full piece here: https://lnkd.in/g-XgS4Xw

It is a bit late, but I am super excited to share that our work, “AutoCircuit: Agentic Pareto Front Exploration for Analog Circuit Designs,” has been accepted as a poster at the ICML 2026 Workshop on AI as a Tool for Mathematics, Computer Science, and Machine Learning (AI4Research), taking place on July 10 in Seoul, South Korea.   In this work, we explore how an LLM can orchestrate optimization tools with complementary characteristics to improve multi-objective analog circuit design under a fixed time budget. Huge thanks to my amazing collaborators Rajath Salegame, Lars Holdijk, and Jan Ole Ernst for making this work possible. This is just one of many exciting research directions we’re exploring at Normal Computing. I’m looking forward to sharing more of what we’re building in the months ahead. Paper (OpenReview): https://lnkd.in/gcgQ7iAY Poster: Attached below. If you’ll be at ICML, I’d love to chat!

Jin Lim joined Normal Computing this week as Vice President, Global Strategic Growth & APAC. Most recently, Jin served as SVP of SK Group's AI data center initiative, where he led both the technology and business strategy behind next-generation AI infrastructure. Earlier in his career, he built and scaled organizations at SK hynix, Samsung, Oracle, Couchbase, and EMC, spanning technical and commercial leadership roles across the full stack of modern compute. At Normal, Jin will lead and expand our top five global Tier-1 semiconductor accounts while owning the three-year APAC P&L, including revenue, margin, and operating discipline. He will build and scale our commercial engagement team across the region and synchronize it with our forward-deployed engineering model we have put in place. Jin's ability to operate across both the technical and commercial dimensions of our customer relationships will be critical as we scale. "Jin brings a rare combination in deep tech; the experience building transformative R&D solutions as a seasoned semiconductor veteran, and the track record of someone who has scaled commercial organizations at the highest level. I look forward to working with Jin to support our closest partnerships in physics-based computing and beyond." – Faris Sbahi, CEO & Co-Founder. We are excited to welcome Jin to the Normal team.

Tim Scarfe of Machine Learning Street Talk (MLST) sat down in Zurich with Thomas Dybdahl Ahle, who leads AI & ML at Normal Computing, on what it takes to bring AI into chip design, and the question underneath it: when an agent produces a design, a proof, or a program, how do you know it is correct? They cover a wide range: the open-source Verilog simulator Thomas and his team built with AI agents, "understanding debt," auto-formalization and the AlphaProof approach applied to chip specs, and our thermodynamic computing work. Watch on YouTube: https://lnkd.in/en47uc2r Listen on Spotify: https://lnkd.in/e55n-kv6 And Thomas Dybdahl Ahle is hiring at https://lnkd.in/gfZ8i7YB if you're interested in applying AI to some of the most critical challenges in silicon.

Solving AI's energy demand is a complex challenge that spans the entire stack, from generation and grid connection to power delivery, cooling, and how efficiently the compute itself runs. Normal is focused on the compute layer: how much intelligence each watt buys. At the Waddesdon Transition Forum, our CEO Faris Sbahi joined the panel "Building the Infrastructure Behind the AI Revolution," alongside James Tyler of Equinix and Jill Macari of VEIR, with James Lockyer of Microsoft's Climate Innovation Fund moderating. The Forum is a gathering of senior investors, policymakers, and energy operators, with Secretary John Kerry as Co-Executive Chair. Thank you to Nadav Steinmetz, Dor Bershadsky, Climate First and Galvanize for including Normal Computing.

Today at the American Swiss Foundation Leadership Summit in Rüschlikon, Normal Computing joined the panel "Powering the AI Era: Meeting the Demand for Compute." Energy challenges are one of the fundamental constraints limiting AI growth, and this discussion explored the energy, compute, and investment stack being built to address it. Normal's CBO Craig Churchill moderated the session with our CEO Faris Sbahi, David Livingston of Galvanize, and Alex Turnbull of Opticore sharing their perspectives on this defining infrastructure challenge of the decade. The panel is part of a broader program bringing together leaders across technology, business, research, academia, and government, including Ambassador Callista Gingrich, former Secretary of Labor Tom Perez, and Senator Rob Portman, to address the pressing issues shaping US-Swiss relations. We're grateful to Vanessa E. B., Caroline Jablow, Emma Carter, and the rest of the ASF team for convening a substantive and well-curated program, and Swiss Re for hosting at the Centre for Global Dialogue.

Sampling is foundational to the thermodynamic hardware we're building and the applied research driving it. Normal Computing is cofunding the second Newcastle Non-Equilibrium Sampling Workshop alongside the Prob_AI Research Hub, bringing together researchers in modern sampling and adjacent areas at Newcastle University from September 2-4. Sam Duffield from our team will be presenting, joined by speakers from Imperial, Warwick, Edinburgh, UCL, Bristol, Lancaster, Cambridge, and others. Pre-registration is open, and the full program and speaker list are on the workshop site: https://lnkd.in/eP8fG7qp

Ed Brandler is joining Normal Computing as Deployment Strategist Lead. We are scaling Normal EDA into production deployments with the world's leading semiconductor companies. Ed will build and lead the Deployment Strategist team to forward-deploy directly with our customers, applying frontier technology to deliver outcomes in the field. Ed began his career as an officer in the British Army before spending more than a decade building and deploying technology alongside customers across Europe, North America, and the Middle East. At Palantir, he was a Deployment Strategist, Commercial Lead, and Country Head, leading complex deployments and embedded engineering teams across government and commercial sectors. He went on to co-found Extrasensory AI, working on neuro-symbolic systems and production AI, and the Sequoia-backed fintech Two. Ed is based in London and we are hiring the founding team now, with roles open across Seoul, New York, London, and San Francisco.

Current approaches involve multiple tools, vendors, designs, data formats, and abstractions. Can agents really use them all? By Brian Bailey. https://lnkd.in/gnPgVspJ #AgenticAI #EDA #EDAmethodologies #HLS #semiconductor Badarinath Kommandur Cadence Doyun Kim Normal Computing Sathishkumar Balasubramanian Siemens Digital Industries Software Shelly Henry Moores Lab AI IC Manage Jeff Tharp Synopsys Inc Simon Davidmann University of Southampton Arvind S. Olivera Stojanovic Vtool - Smart Verification