As part of SimScale’s Engineering Leaders Webinar Series in collaboration with engineering.com, SimScale and Accenture brought together two expert perspectives to unpack the state of Engineering AI heading into 2026.
Featuring David Heiny, Co-founder and CEO of SimScale, and Vlad Ladichev, Industrial AI Lead at Accenture, the session tackled the question on every engineering leader’s mind: why are so many AI programs stuck in pilot purgatory, and what does it actually take to break through?
The webinar coincided with the launch of SimScale’s State of Engineering AI 2026 Report, a survey of 350 CTOs and VP Engineers across the US, UK, and Germany. If you missed the live session, here are the top five highlights.
On-Demand Webinar
Watch the full session on demand, including live demos of SimScale’s Engineering AI agent registry and Physics AI in action.
1. The Shift Is Real: AI Pilot Purgatory Is Starting to Break
The most significant finding from the 2026 report is a measurable step-change in Engineering AI maturity. In our last survey in 2025, most survey respondents were already in agreement: AI will certainly play a major role in hardware engineering, and the opportunity – and the expectation – was significant. But very few organizations had turned this vision into reality, with many having not even started to explore what was possible.
2026 is different. For the first time, a statistically significant group of respondents (the AI leaders) reported actual impact from AI systems running in production. Organizations experimenting with AI pilots doubled year-over-year.
As David Heiny put it during the session: “Everybody is clear that a similar level of transformation will happen in hardware engineering as in software. What nobody is yet entirely clear on is how and what it will look like.”
Key takeaway: The question has shifted from “how and when should we start?” to “how do we move from pilots to scaled production value?” The leaders who’ve made that leap are pulling ahead, and their results are now showing up in survey data for the first time.
2. Design Space Exploration: AI Leaders Are Evaluating 2-4x More Design Variants
One of the most concrete findings in the report: engineering teams using AI-enabled workflows are evaluating 2 to 4 times as many design variants per program compared to those using conventional approaches. This showed up in survey data for the first time in 2026, validating what many in the field had experienced anecdotally.
David explained the mechanism: simulation teams typically face a lead time of 4-8 weeks between when a simulation is requested and when an engineer picks it up. When Engineering AI agents handle standard simulation workflows autonomously, that wait time is eliminated, which directly drives broader design exploration.
“There’s an endless amount of simulations that never get run because the simulation team never gets around to them,” David noted. “By cutting out that wait time, you see a jump in the ability to explore designs.”
Key takeaway: The value of Engineering AI isn’t just faster compute. It’s the elimination of bottlenecks across the entire design-to-simulation (and back again) cycle. Broader design space exploration is the outcome, and it’s now measurable.
3. Commercial Impact Is No Longer Just a Promise
The 2026 report surfaced something genuinely new: AI leaders aren’t just reporting technical process improvements; they’re reporting commercial business impact. This is particularly visible in scenarios where engineering teams must respond to RFQs (Requests for Quotation) under tight deadlines.
Vlad Ladichev, drawing on Accenture’s industrial AI practice, framed it clearly: “When you can evaluate more design options in an RFQ response window, you’re not just faster; you protect your margins and make sure what you’re supplying will actually work. Speed becomes your market capture weapon.”
The implication for engineering leaders: AI-accelerated design exploration reduces the risk of underbidding, improves the quality of engineering proposals, and directly impacts revenue. It’s a key insight into how the organisations with the most mature AI programs make it an intrinsic part of engineering workflows to deliver into top-line commercial outcomes.
Key takeaway: For leaders building the business case for AI investment, this is the signal to anchor on. Mature Engineering AI programs are now delivering commercial outcomes, not just efficiency gains.
4. The Real Enablers Haven’t Changed, But More Leaders Are Finally Acting
The top enablers of successful Engineering AI programs remained consistent year-over-year: data governance, cloud-native infrastructure, leadership buy-in, and organizational readiness. What changed is that more organizations are now actively prioritizing them rather than treating them as future prerequisites.
Vlad offered a practical framework on the data question, one that stops many companies before they start: “You need to know what data you actually have, who can access it, and what you’re allowed to do with it. Answer those three questions and you’ve won 70% of the battle.”
He pushed back against waiting for a perfect data foundation, seeing it as a distraction from the critical objective of finding out what works and what doesn’t: “Companies bunker data for decades, waiting for some magical technology to come along and extract the value. Don’t. Start with one high-impact use case that doesn’t need perfect data, build your AI muscle, and grow horizontally from there.”
On infrastructure, David pointed to SimScale’s AI-native cloud simulation platform as the underlying enabler: “It’s not enough to give engineers faster compute. You need a mechanism that scales simulation expertise across the organization with broad use and central governance. Cloud-native architecture makes that possible, and it means Engineering AI doesn’t require a big IT project to get started.”
Key takeaway: The blockers are not new. The differentiating factor is whether organizations have the leadership buy-in and infrastructure to start acting now, not waiting for perfect conditions that never arrive.
5. From Co-pilot to Agent: Agentic Engineering Is No Longer Just Theory
A year ago, “AI assistant” and “co-pilot” were the dominant mental models for how AI can help engineers. In 2026, the conversation has moved to agents: AI systems that don’t just advise but take autonomous action. The report shows that some engineering organizations are beginning to permit agents under human guidance for specific workflows, though most remain in assisted/advisory mode.
Both speakers converged on why agentic AI is harder to deploy in engineering than in software: code has a compiler. When software agents write and test code, there’s an immediate binary feedback signal; it either compiles or it doesn’t, and then coding agents can directly test the finished product themselves. It is much harder to close that loop in hardware engineering, but there are many steps and inner loops in the product development process where AI agents can drive significant acceleration – and the leading teams are fast discovering them.
David demonstrated SimScale’s Engineering AI agent registry during the session: an agent that autonomously sets up and runs simulation workflows based on plain-text instructions and example simulations, without a training step. Combined with Physics AI (which predicts simulation results in seconds by training deep learning surrogate models on prior runs), the demo showed what production-ready agentic Engineering AI looks like today.
Vlad shared Accenture’s real-world application: an agentic system for battery frame design that reads requirements, iterates overnight across hundreds of design variants, and surfaces the passing designs to engineers in the morning, saving up to 40% of time in the trial-and-error phase.
Key takeaway: The move from co-pilot to agent is underway in engineering, but it requires trustworthy, closed-loop systems, not just powerful models. The organizations that get this right will define the next phase of Engineering AI adoption.
Final Thoughts
This session brought together quantitative survey data and real-world practitioner experience from two organizations working at the frontier of Engineering AI. The through-line is clear: the question is no longer whether Engineering AI will transform hardware engineering. It’s whether your organization will be in the group driving that transformation or catching up.
For teams still working through their first use case, David and Vlad’s advice aligned: Don’t wait for perfect data, perfect infrastructure, or perfect organizational alignment. Start with a concrete, high-value workflow, get an early win, and build from there. The leaders pulling ahead didn’t wait either.
To dig deeper into the survey findings, including a detailed breakdown of the AI leaders cohort and what separates them from the rest, download the full report.
