Neither of us made it to Long Beach for this year’s Design Automation Conference, but Paul has been reviewing the announcements and subsequent commentary. I sat down with him to get his take on what matters for our customers. 

Pete Davy: So, DAC 2026. You weren’t there and I wasn’t there. What’s your first impression from everything you’ve read? 

Paul Double: Honestly? A bit of “here we go again” at first. Every DAC for the past few years has  had an AI headline attached to it. Once you get beyond the announcements and look at what Siemens and others are productising and preparing to release, however, there is a real shift underneath it. It is no longer simply “AI helps you write a script”. These are AI agents that can reason about an objective, use the actual EDA tools, examine the results and decide what to do next. That is a different proposition.

Pete: Is that just marketing language for “more automation”, though? 

Paul: That is a fair question and one I asked myself. What convinced me that it is more than that is the way Siemens is pairing AI with deterministic engineering engines. These include Calibre for physical verification, alongside expanded work with NVIDIA intended to let agents check their decisions against physics-based tools as they proceed. The AI proposes an action, but an established engineering tool still has to confirm that the result is valid. That is a more credible way of addressing the trust problem. 

Pete: What has landed best with the kinds of customers we deal with, such as smaller custom IC houses and analogue and mixed-signal teams? 

Paul: Solido Layout Analyzer, without question. It is aimed squarely at post-layout debugging, helping engineers trace why a layout behaves differently from the simulation. That would normally require someone to cross-reference numerous views and reports manually. 

The software links layout measurements directly to circuit behaviour. With the Fuse agent layered on top, engineers can ask questions in plain language instead of searching through reports themselves. STMicroelectronics said the approach could cut weeks from debugging complex non-volatile memory blocks, while noting that it still planned to test and validate the benefits through ongoing design work. If the results bear that out, it could significantly shorten project schedules.

Pete: Anything on the digital and DRC side? A lot of our conversations are still about physical verification headaches.

Paul: Yes. Calibre Vision AI, which Siemens introduced in 2025 and continued to develop during 2026, uses machine learning to cluster, prioritise and visualise large volumes of DRC violations. 

Siemens’ more direct digital verification news at DAC concerned the Questa One Agentic Toolkit and its integration with NVIDIA’s Nemotron reasoning technology. That extends the agentic approach into verification workflows rather than limiting it to physical verification and custom IC design. 

Pete: You mentioned that Siemens announced a couple of planned acquisitions around DAC too. 

Paul: Yes, and I think they are more telling than many of the product demonstrations. Siemens announced agreements to acquire Precision Innovations and Defacto Technologies. 

Precision Innovations provides early chip-planning technology built on OpenROAD, allowing teams to examine architectural trade-offs before becoming committed to a particular direction. Defacto Technologies, based in Grenoble, specialises in automated SoC design creation, assembly and integration. 

Both moves push automation earlier into the flow. The logic is straightforward: the earlier you identify a poor architectural decision, the less expensive it is to reverse. Defacto’s location in Grenoble also makes it particularly relevant to our European customers. 

Pete: Was there anything from the panel discussions that stood out, rather than from the product side? 

Paul: There was a useful debate about build versus buy and, ultimately, who owns the intelligence behind these workflows. As process and design data become a greater source of competitive advantage, teams must think carefully about how much access third-party AI should have to their intellectual property.

For many of our customers, particularly smaller and IP-sensitive design houses, that may be a more pressing question than whether the AI itself is capable enough. Some organisations will build their own intelligence layer around established EDA engines, while others will rely more heavily on vendor-provided agents. Many will probably use a mixture of the two. 

Pete: If you had to sum it up for someone running a small ASIC team who does not have time to read everything, what should they take from it? 

Paul: The engineers are still in charge. None of this replaces engineering judgement. Someone still has to define the objective, evaluate the trade-offs and confirm that the result is right for the product. 

What is changing is the amount of repetitive investigation and tool operation that engineers must perform themselves. For a smaller team without much spare capacity, that could be genuinely useful. It means the engineering expertise available within the team can cover more ground. 

It is also important to distinguish what is available now from what has been announced for future releases. Some of the underlying tools are already established, while several of the expanded agentic capabilities presented at DAC are still forthcoming. 

Pete: Where do we fit into that from EDA Solutions’ perspective? 

Paul: In much the same place as always. We are here to determine where the technology solves a genuine problem within a customer’s flow, whether that involves simulation capacity, post-layout debugging, design closure or verification. 

If someone reading this wants to discuss where any of these developments might apply to their work, that is exactly the conversation we want to have. 

Pete: Paul, thanks. It is good to get the full picture without either of us having to queue for a badge in Long Beach. 

Paul: Next year, maybe. 

If any of this raises questions about your own flow, get in touch with the team at EDA Solutions.