Inside Synopsys Startup & VC Connect: Building the Future of AI Silicon

“The next NVIDIA could be in this room.” 

That’s how Antonio Varas, SVP and chief strategy officer at Synopsys greeted the 130 attendees at the inaugural Synopsys Startup Program event last month at the Synopsys, Inc. headquarters in Sunnyvale. At Synopsys Startup & VC Connect, industry leaders, emerging startups, and venture capitalists came together to explore what it truly takes to bring next-generation silicon to life in an era defined by artificial intelligence (AI) and the challenges AI brings to scale and technology integration.

From architecture to investment, a clear message emerged: success in today’s semiconductor landscape depends not just on breakthrough ideas, but on speed, ecosystem collaboration, and the ability to de-risk every stage of development. 

Synopsys leaders, accomplished startup founders, and seasoned venture capitalists took the stage to share insights on what it takes to succeed as a silicon startup in today’s landscape.

The Synopsys Startup Program is aimed to support startups, no matter the stage they are in. “We want to hear, listen, and learn about how Synopsys can be more accessible and meet startups where you are,” explained Varas.

The AI semiconductor market is projected to reach $1.2 trillion by 2030, with the CPU segment emerging as a key growth driver. Meanwhile, investment in custom silicon architecture is growing faster than general-purpose accelerators — a clear signal of the industry's shift toward purpose-built silicon, optimized for specific AI workloads.

Synopsys leaders outlined a clear roadmap for how startups can better navigate the journey from concept to first silicon. A key focus was reducing risk early in the design process through silicon-proven interface IP and “shift-left” methodologies, giving teams access to validated building blocks that are already optimized for performance, power, and interoperability. By leveraging proven IP rather than starting from scratch, startups can dramatically compress development timelines while minimizing the risk of costly redesigns later in the process. Paired with shift-left approaches, which address critical design considerations, verification, testing, and problem-solving earlier in the development process, this enables faster software bring-up and design convergence, plus more predictable paths to tape-out.

“Almost every customer tells me the same thing — 'Just give us IP that works.' Behind that ask is a team chasing a market window, a runway, a dream. The industry has moved from IP blocks to pre-validated subsystems, because in AI silicon, integration risk is the difference between shipping and stalling. Our job is simple: take that risk off your shoulders, so you can focus on differentiating your product, not on basic integration and bring-up.” says Neeraj Paliwal, senior vice president of product management at Synopsys.

Together, these approaches give startups a more predictable, lower-risk path to first silicon—enabling them to innovate faster, scale with confidence, and bring differentiated products to market with greater efficiency.

Through cloud-scale emulation infrastructure like Synopsys ZeBu Cloud, startups can tap into on-demand infrastructure that accelerates hardware-software integration while preserving the flexibility to scale alongside evolving designs. This shift is critical: Infrastructure is no longer a bottleneck, but an enabler of speed and innovation for even the smallest teams.

“In today’s AI-driven environment, startups play a pivotal role in shaping the next generation of AI- and software-driven products. Their speed, creativity, and willingness to challenge convention are essential drivers of meaningful innovation. At Synopsys, our system verification and validation solutions empower startups to move faster and with greater confidence, enabling them to bring complex ideas to market efficiently. With ZeBu Cloud, we are also making hardware-assisted verification far more accessible—providing an easy entry point for startups to leverage powerful capabilities without the traditional barriers to adoption,” says Tom De Schutter, senior vice president of product management at Synopsys.

Building on this foundation, Synopsys highlighted new advancements in its hardware-assisted verification (HAV) portfolio, introducing a software-defined approach purpose-built for the escalating complexity of AI silicon. With up to 2x gains in performance and capacity, , these innovations expand emulation and prototyping at the scale required for multi-die, AI-driven architectures. Ultimately, HAV is positioned as a cornerstone for achieving first-time-right silicon — empowering teams to validate complex systems more efficiently, scale verification workloads with confidence, and meet increasingly aggressive time-to-market demands.

Combined with Synopsys proven IP and HAV optimized interface solutions for that IP, this creates a powerful multiplier effect in which teams can move faster not only because they’re building on trusted foundations, but because they’re able to iterate, validate, and refine designs in parallel. The result is a more efficient path to silicon that enables startups to focus on differentiation while relying on a mature ecosystem to handle complexity.

Equally important was the discussion of the growing role of AI within the design process itself, particularly as traditional EDA approaches struggle to keep pace with the exponential complexity of AI-driven chips. Synopsys leaders emphasized that AI-driven EDA is not simply an incremental improvement, but a fundamental shift in how chips are designed and optimized — automating traditionally manual tasks, accelerating verification cycles, and enabling engineers to explore far larger design spaces with greater confidence. When combined with multiphysics simulation from Ansys, part of Synopsys, teams can model interactions across electrical, thermal, and mechanical domains earlier in the process, reducing downstream surprises and enabling more informed design decisions from the outset.

“Startups cannot afford multi-year learning cycles to prove an architecture will scale. The advantage now comes from moving beyond overdesign to co-design — uniting software and hardware, electronics and physics, in a single intelligent workflow. With Synopsys EDA, Multiphysics Fusion, AgentEngineer, and cloud-native delivery, a ten-person team can execute with the speed, insight, and confidence that once required an organization ten times the size. That is the real unlock,” says Sanjay Bali, senior vice president of product management at Synopsys.

Generative AI is further extending these capabilities by helping teams rapidly generate, evaluate, and refine design options, unlocking new levels of productivity at a time when engineering resources are stretched thin. Together, these technologies are enabling a more predictive and data-driven approach to chip development, helping startups manage soaring complexity, improve design quality, and increase the likelihood of first-pass silicon success. The net effect is a redefinition of what’s possible for emerging companies, enabling them to accelerate innovation cycles while lowering traditional barriers to entry, all without compromising on ambition or performance.

The startup panel featured leaders from Rapidus, DensityAI, and Cerebras, who offered a candid look at what it takes to bring AI silicon from concept to production. A consistent theme was the importance of grounding innovation in real-world application early. Panelists emphasized starting with the software and workload first, before moving to silicon design. By deeply understanding the problem to be solved upfront, teams can make more informed architectural decisions and avoid costly pivots later in the process. This “software-first” mindset reflects a broader shift in the industry, where success is increasingly defined by how well silicon maps to specific AI workloads.

“Start with software first, not the silicon,” advised Bejamin Floering, head of engineering at Density AI.

Equally important was the focus on execution and the often underestimated “last mile” to production. Leaders highlighted challenges such as design closure, foundry readiness, and system-level validation as critical inflection points where many startups struggle. In this environment, partnerships become a key success factor. Choosing the right design and ecosystem partners, and investing in those relationships early, was noted as essential to navigating complexity and reducing risk. The takeaway was clear: Breakthrough ideas alone are not enough. Startups that succeed are those that combine strong technical vision with disciplined execution, trusted collaboration, and a deep understanding of the end application from day one.

The venture capital panel offered a complementary perspective, focused not just on building AI silicon, but on what it takes to fund and scale it successfully. Investors from leading firms Celesta, Lam Research, and Plug and Play emphasized that, in today’s environment, technical ambition must be matched with strategic clarity and disciplined execution. Differentiation remains critical, but it must be rooted in a clearly defined problem and a compelling advantage over existing solutions. With development costs for advanced AI chips reaching into the billions, venture decisions are increasingly shaped by a startup’s ability to demonstrate not only innovation, but also capital efficiency, execution speed, and a credible path to market. 

Just as important, investors highlighted the growing role of ecosystem alignment in de-risking investment. Startups that engage early with strong partners — whether in design, IP, or manufacturing — are better positioned to accelerate development and validate their approach. The idea of having a “built-in” partner or early customer resonated strongly, reinforcing that success in AI silicon is rarely achieved in isolation. Ultimately, the panel underscored a shift in how venture capital evaluates opportunity in semiconductors. VCs are not simply betting on breakthrough ideas, but on teams that can execute within a complex, capital-intensive ecosystem, and bring innovative architectures all the way to production.