A Year of Careful Acceleration
There is something almost tender about watching an entire industry try to slow down and get things right, even while it races forward. That is the mood running through chip design news this year, as engineers, researchers, and toolmakers quietly rethink not just how fast a chip can be built, but how confidently it can be trusted. Two currents are shaping this shift more than any other: the rise of AI agents that verify their own work, and a growing, generous movement toward open source silicon. Neither story is loud on its own, but together they say something meaningful about where the industry is heading.
Long Beach Hosts a Turning Point
For the first time in its sixty-three-year history, DAC, The Chips to Systems Conference, moved to Long Beach, California, drawing record participation from across the design ecosystem. The gathering reflected an industry at an inflection point, with growth across both research and engineering tracks and an unmistakable focus on AI enabled design methods. More than one hundred and twenty exhibitors filled the show floor, spanning electronic design automation, intellectual property, and semiconductor tools. The choice of a new home for such a long running event felt symbolic, as though the field itself needed more room to hold everything happening within it.
Why Verification Has Always Been the Hard Part
Long before agents and automation entered the conversation, verification was already known as one of the most demanding, unglamorous stages of chip design, where patience mattered more than raw cleverness. A few reasons explain why:
- Engineers could spend months tracing through simulations to catch a design’s true behavior
- Rare edge cases might only surface under narrow, unlucky conditions
- A single missed flaw could mean a costly redesign or a failed chip already in manufacturing
- This slow, careful labor is exactly what this year’s AI tools are now being asked to help carry
Synopsys Teaches Its Agents to Check Their Own Work
At DAC 2026, Synopsys introduced a fully autonomous design verification agent capable of orchestrating an entire chip verification cycle on its own, a shift the company says can bring validated designs many times faster than before. Built alongside NVIDIA’s accelerated computing platform, the workflow pairs domain specific agents with broader reasoning systems to identify design failures and speed up root cause analysis. Synopsys also unveiled its first fully autonomous workflow for electronics system design, extending the same philosophy beyond the chip itself. The company frames this less as replacing engineers and more as giving them room to focus where their judgment matters most.
Cadence Reaches What It Calls Full Autonomy
Cadence has spent the past year mapping out five levels of AI maturity in chip design, moving from simple optimization tools up through conversational assistance, complex reasoning, and coordinated agent workflows. At the Computex event this year, the company’s ChipStack AI Super Agent reportedly reached the fifth and final level, described as full autonomy from specification through verification with minimal human involvement. At DAC 2026, Cadence extended that ambition further with its AuraStack AI Super Agent, aimed at printed circuit board and advanced packaging design. Taken together, the announcements complete an agent portfolio that now spans nearly the entire electronic system design flow.
Siemens Adds Self-Verifying Agents to the Mix
Not to be left out of the moment, Siemens introduced self-verifying agentic workflows within its Fuse EDA AI Agent system, built on the same broader agentic AI foundation as its two larger rivals. The idea of a self-verifying agent captures something quietly important about this whole wave of tools: the goal is not simply speed, but confidence that the work an AI system produces can check itself before a human ever needs to step in. Analysts watching the conference noted that having all three major EDA companies move toward long running autonomy at the same moment marked a genuine turning point, rather than a coincidence of timing.
NVIDIA’s Quiet Role Underneath It All
Much of this year’s agentic momentum runs on infrastructure that rarely gets top billing. NVIDIA expanded its Agent Toolkit with new physics and computing libraries this year, while also disclosing that it has begun deploying its own processors across its internal electronic design automation workflows. That detail is a small but telling one: even the company building tools for others to design chips faster is now using AI agents to design its own. It is a reminder that the shift toward autonomous verification is not a marketing idea sitting on top of the industry, but something being tested from the inside out.
Startups Racing to Automate the Bottleneck
Alongside the established EDA giants, a newer generation of companies is pushing agentic AI into chip design from a different angle. Firms such as Agentrys, Cognichip, and ChipAgents are each carving out a distinct approach, whether through workflow orchestration, physics informed foundation models, or vertically integrated platforms built on proprietary models of their own. Verification remains their shared starting point, since it continues to be one of the most time consuming bottlenecks in getting a design from concept to working silicon. Whether any of these smaller players can match the scale of Synopsys, Cadence, or Siemens remains an open question, but their energy is undeniable.
ChipAgents’ Rapid Growth Tells Its Own Story
Few numbers capture this year’s momentum as clearly as ChipAgents’ funding round. The company expanded its Series A financing to one hundred and thirty-four million dollars, following sixfold growth in annual recurring revenue during the first half of 2026 and deployments across more than one hundred and twenty semiconductor companies, including well known names like MediaTek and Micron. Its founder has described the goal as moving customers beyond simple AI assistance and toward agents capable of doing meaningful engineering work on their own. Independent evaluations this year found the platform outperforming generic AI agents by a wide margin on commercial scale intellectual property, including bus fabrics and RISC-V cores.
Executives Gather to Ask the Hard Questions
Not everyone in the industry is simply celebrating. In the spring, the Electronic System Design Alliance hosted its annual Executive Outlook panel, gathering EDA, agentic AI, and intellectual property leaders at Cadence’s own headquarters to discuss how agentic AI will genuinely change chip design and verification, rather than simply promise to. Sessions like this matter because they slow the conversation down, inviting people who build these tools every day to be honest about their limits as well as their promise. It is a small but healthy sign that the industry is not letting its own excitement go unquestioned.
The Open Silicon Movement Keeps Growing
While AI dominates the headlines, a second and gentler story has been unfolding alongside it: the steady maturing of open source chip design. Where proprietary tools and licensing fees once locked smaller teams and universities out of serious semiconductor work, an expanding ecosystem of freely available design software, verification frameworks, and processor blueprints is quietly lowering that barrier. This movement does not chase the same speed records as agentic AI, but it carries its own kind of significance, widening who gets to participate in building the technology that underpins nearly everything else. In many ways, it is the more democratic half of this year’s chip design story.
A New Chapter for the CHIPS Alliance
The CHIPS Alliance, a central hub for open source silicon projects, described 2026 as a year of transition, moving from an early incubation phase toward a more mature, community driven ecosystem centered on trust and technical rigor. The organization is also shifting its own leadership model, with longtime general manager Rob Mains stepping back in favor of a community driven approach meant to keep the group financially and operationally sustainable over the long run. Established projects like Caliptra continue to anchor the portfolio, even as newer workgroups form around emerging needs. It is the kind of quiet institutional change that rarely makes headlines but shapes what gets built for years afterward.
Exploring Open-Source AI for Chip Design
Perhaps the most forward looking piece of the CHIPS Alliance’s plans for this year is an early exploration into forming an AI and machine learning workgroup dedicated to open source frameworks for agentic chip design tools, potentially built in partnership with several American universities. There is also discussion underway about bringing the Coral NPU project into the Alliance, which would offer a complete, openly available platform for edge AI hardware. Both efforts suggest that the open source community does not intend to simply watch the big commercial AI verification tools from the sidelines. Instead, it hopes to build its own version of the same idea, openly and collaboratively.
RISC-V’s Steady Rise Around the World
Much of the open source hardware conversation still centers on RISC-V, the freely licensed processor architecture that continues gaining ground against long established, proprietary alternatives. Unlike closed architectures controlled by a single company, RISC-V can be used, modified, and built upon by anyone without licensing fees, which has made it especially attractive to startups, universities, and countries seeking more independent semiconductor supply chains. Commentators this year have described the moment as a particularly meaningful one for regions such as India, where open architectures and open source tools are seen as a genuine opportunity to become relevant in global semiconductor design rather than simply consuming chips designed elsewhere.
China’s Xiangshan Processor and openRuyi
Nowhere is that RISC-V momentum clearer than in China, where researchers at the Chinese Academy of Sciences say they have built the most capable open-source RISC-V processor in the world, known as Xiangshan. The project sits alongside openRuyi, a native operating system designed specifically to run on that same open hardware core. Officials describe the effort as part of a broader push toward computing independence, reducing reliance on architectures controlled abroad. With more than six hundred hardware researchers and four hundred software researchers reportedly involved, it is one of the largest coordinated open-source silicon efforts anywhere in the world today.
A Student Team in Bangladesh Builds Its Own Chip
Far from the scale of a national research institute, a university team at the Bangladesh University of Engineering and Technology quietly accomplished something worth pausing on this year. Using only free, openly available design tools, the researchers designed, verified, and taped out a complete thirty-two bit RISC-V processor as a full system on chip, without relying on any proprietary electronic design automation software. The project shows how far open tooling has come, allowing an academic team with limited resources to complete work that once required expensive commercial licenses. It is a small story, but a genuinely hopeful one about who gets to learn how chips are actually made.
What Open Tools Mean for Smaller Players
Stories like the one from Bangladesh point toward something larger than any single processor: when design tools are free, the starting line quietly moves closer for everyone.This change can be seen in several practical ways.
- Universities and student teams can design, verify, and tape out real chips without paying for costly commercial licenses
- Startups in developing regions can attempt semiconductor work once reserved for well funded companies
- Small research groups gain room to experiment and learn, even without deep institutional backing
- Commercial EDA tools remain dominant for now, but open tools are steadily narrowing the gap
The Business Behind the Numbers
Underneath all of this innovation sits a semiconductor market that is, by most measures, thriving. According to industry tracking this year, global semiconductor sales reached roughly one hundred twenty billion dollars in a single month, more than double what they were during the same period the previous year. Government backed research awards continue flowing toward advanced compute supply chains, supporting technologies tied to artificial intelligence and healthcare alike. None of the design breakthroughs described here happen in a vacuum; they are unfolding against a backdrop of real financial momentum, which helps explain why so many companies are willing to invest so heavily in both agentic AI and open ecosystems at once.
A Gentle Closing Thought
Taken all together, this year’s chip design news tells a story that is less about speed for its own sake and more about care taking new forms. Agentic AI is teaching itself to double check its own conclusions before handing them to a human. Open source communities are teaching entire regions and universities how to build chips that once felt permanently out of reach. Neither trend erases the need for patient, skilled engineers; if anything, both seem to be making more room for that patience to matter. It is a hopeful, unusually collaborative moment for an industry that spends most of its time working quietly beneath everything else we use.
