Siemens EDA details Questa One's three-pillar AI update at DAC 2026

Siemens EDA used DAC 2026 in Long Beach, California - running 26-29 July 2026 - to brief analysts on the latest state of Questa One, its AI-infused smart verification platform^[1]. Abhi Kolpekwar, VP and General Manager of Digital Verification Technologies at Siemens EDA, framed the update around a single pressure point: complex SoC and chiplet designs now routinely require three to four years of new product development, and traditional EDA tools are no longer keeping pace with the scale of those designs^[1].

Three features driving the update

Siemens structured the Questa One update around three capabilities^[1]:

• Faster engines - raw simulation and formal performance improvements across the core Questa One toolset.
• Bidirectional inter-engine communication - engines now exchange data through micro-services, a change that also lets AI agents query and drive individual engines directly^[1].
• Model Context Protocol (MCP) support - the micro-service layer enabled Siemens to expose its engines via MCP, giving external agentic frameworks a standardised interface into Questa One's verification state^[1].

The third point matters for interoperability. Siemens describes its methodology as LLM-agnostic, letting engineers choose their own reasoning models, while noting that NVIDIA NIM and Nemotron are frequently used for high-performance reasoning tasks^[1]. The architecture is also compatible with mainstream AI coding environments - including GitHub Copilot, Claude Code, Cursor, and VS Code - so teams are not forced into a single toolchain^[1].

The Agentic Toolkit and what it covers

The Questa One Agentic Toolkit, announced on 27 February 2026, is the delivery vehicle for these agentic workflows^[1]. It was announced on 27 February 2026 and integrates with Siemens' Fuse EDA AI system as its preferred orchestration layer, while remaining open to other agentic platforms. The toolkit provides domain-specific intelligence that plugs into Siemens' Coding Agent and Fuse Agent orchestrators^[1].

At DAC, Kolpekwar confirmed that the toolkit ships with five agents covering the most time-intensive steps in digital verification^[1]:

• RTL Code Agent - converts natural-language descriptions into synthesisable RTL and checks for coding violations.
• Lint Agent - configures lint analysis, flags design errors, and offers automated fixes or waivers.
• CDC Agent - runs clock domain crossing verification and suggests configuration fine-tuning based on results.
• Verification Planning Agent - analyses design specifications to generate test plans.
• Debug Agent - assists with root-cause analysis across simulation runs.

Verification consumes up to 70 percent of total design time and complexity, according to Siemens, which is the productivity gap the toolkit is designed to close^[1].

Where the human stays in the loop

Siemens was explicit at DAC that Questa One is not positioned as a path to fully autonomous IC verification. The tools are designed to amplify engineering judgement rather than replace it, with customer-defined governance boundaries controlling how far agents act without approval^[1]. MediaTek was cited as a customer that has used Questa One to bring new engineers up to speed faster on RTL methodology, compressing what would typically be a multi-day learning curve^[1].

The first generation of AI-based EDA tools operated as isolated point tools; in 2026, the technology is being used to orchestrate multiple EDA tools toward higher-level objectives^[1].

The signal to watch is how broadly Siemens can extend MCP support across the rest of its EDA portfolio beyond Questa One. If the inter-engine micro-service layer proves stable at production scale, it becomes the plumbing for a much wider set of agentic workflows - and a potential differentiator against point-solution verification startups that lack native engine access.