2026 Outlook with Ying J Chen of S2C

I’m Ying J Chen, VP of S2C. S2C is a leading global supplier of FPGA prototyping solutions for advanced SoC and ASIC designs, holding the second largest share of the global prototyping market. Founded in 2003, the company has supported more than 600 customers, including 11 of the top 25 semiconductor companies worldwide, with teams and operations across the U.S., Asia, Europe, and ANZ.

What was the most exciting high point of 2025 for your company? 

One of the most exciting highlights of 2025 was seeing our long-term RISC-V investments translate into concrete, system-level results through close collaboration with ecosystem partners.

Working with the Beijing Open Source Chip Research Institute (BOSC), we completed key system validation of multiple generations of the Kunminghu RISC-V processors on our Prodigy S8-100 Logic System. The dual-core Kunminghu V2 RISC-V processor successfully booted GUI-based OpenEuler 24.03 at 50 MHz on S2C’s Prodigy S8-100 Logic System, running real applications such as LibreOffice and the classic DOOM game. We also validated BOSC’s third-generation 16-core Kunminghu processor with a NoC interconnect on two S8-100Q Logic Systems (each with four VP1902 FPGAs), achieving stable timing closure at 13.3 MHz and demonstrating scalability for more complex designs.

In parallel, we demonstrated Andes Technology’s AX45MPV vector processor IP running a live large language model on our Prodigy S8-100 Logic System. Together, these milestones highlighted our ability to support high-performance, multi-core RISC-V systems with real software workloads, reinforcing S2C’s role as a trusted prototyping partner in the RISC-V ecosystem.

What was the biggest challenge your company faced in 2025?

The biggest challenge in 2025 was managing rapid growth in design scale and system complexity, especially for AI-focused SoCs. Customers increasingly need high capacity, fast execution, and deep debug visibility at the same time. Verification has moved well beyond RTL correctness—teams are bringing up full systems with complex software stacks. This puts pressure on infrastructure while increasing sensitivity to cost, deployment effort, and workflow continuity. Tighter schedules and market uncertainty further pushed customers to look for platforms that can adapt across different development phases.

How is your company’s work addressing this challenge? 

A key issue we see is the trade-off between execution speed and debug depth. Traditionally, prototyping and emulation are handled by separate systems, which adds cost and slows iteration.

Our response has been to rethink how these needs are supported across a project’s lifecycle. With OmniDrive, we’re exploring a dual-mode approach built on a shared hardware foundation, allowing teams to use the same platform for fast software bring-up or deeper debug, depending on the stage of development. While the modes aren’t interchangeable at runtime, this approach helps reduce duplicated infrastructure and improve overall price-performance.

This direction is being refined through close collaboration with early customers, where we’re validating that it holds up in real engineering environments, both technically and economically.

What do you think the biggest growth area for 2026 will be, and why? How is your company’s work addressing this growth? 

We see RISC‑V and AI‑silicon as the key growth drivers in 2026. RISC‑V is moving into broader, more customized deployments, while AI workloads are drastically increasing design scale and complexity.

To support this, we provide a verification infrastructure built for scale. Our RTL Compile Flow (RCF) and Incremental Compile Flow (ICF) efficiently handle very large designs and are proven in multi‑FPGA deployments, shifting verification left.

In 2026, we will promote OmniDrive, our next‑generation emulation system. Based on latest FPGA architecture, its “dual‑mode” supports both emulation and prototyping, enabling high‑speed verification and fine‑grained debugging. This significantly reduces customer hardware investment and total cost.

Together, RCF, ICF, and OmniDrive offer a smooth progression from bring‑up to full‑system validation, helping customers scale while controlling both technical risk and infrastructure cost.

What conferences did you attend in 2025 and how was the traffic?

In 2025, we were selective about which conferences we attended and how much value they delivered. We’ve participated in DAC for many years, but this was the first year we chose not to attend. For us, DAC has become less effective in terms of qualified traffic and meaningful technical engagement, especially given how our customer base and focus areas have evolved.

In contrast, RISC-V–focused conferences, including Andes’ RISC-V CON and related ecosystem events, generated much more relevant and application-specific interaction. The audience there was closely aligned with what we’re working on—real systems, software bring-up, and system-level verification—so the conversations were more relevant and more actionable.

We also saw solid results from DVCon events across different regions. While traffic varied by location, the overall quality was strong, particularly among verification engineers and technical decision-makers. These events continue to be valuable for in-depth technical discussions rather than broad marketing exposure.

Overall, in 2025 we saw a clear shift away from large, general-purpose conferences toward more focused, domain-specific events. For us, relevance and engagement mattered far more than raw foot traffic, and that’s increasingly guiding where we invest our time.

Will you participate in conferences in 2026? Same or more as 2025?

We expect to remain active in industry conferences in 2026, at a similar level or slightly higher than in 2025. Our focus will continue to be on technically driven events that support deeper engineering discussion.

As SoC designs continue to scale up in size and complexity, we’ll place more emphasis on demonstrating how OmniDrive, together with our partitioning and RTL compile technologies, helps teams handle larger designs more efficiently. Rather than broad visibility, we prioritize venues where we can show practical system capability and engage with customers facing real scale and integration challenges.

How do customers normally engage with your company?

Customers usually first connect with us through events, press announcements, media coverage, advertising, or organic search. These initial touchpoints often spark technical conversations, such as system capacity, partitioning, or full-system bring-up challenges. From there, we work closely with engineering teams through evaluations, demos, and proof-of-concept projects. Given the scale and complexity of modern SoCs, these relationships tend to be long-term and collaborative, often spanning multiple design iterations. This approach ensures our engagements are practical, technically meaningful, and directly tied to customer needs.

Are you incorporating AI into your products? Is AI affecting the way you develop your products?

Yes, AI is starting to influence both our products and how we engage with customers. We’ve begun applying AI-driven techniques within our tool flows to help shorten compile time, reduce verification cycles, and improve performance efficiency in customer deployments.

At the same time, we’re preparing an AI-enabled knowledge base based on large models to support faster, more consistent technical guidance. While this capability is still being refined, the goal is to improve support efficiency without overcomplicating the engineering workflow.

Overall, we see AI as a practical enabler—used selectively to accelerate development and improve the customer experience in complex SoC and AI-driven designs.

Also Read:

S2C, MachineWare, and Andes Introduce RISC-V Co-Emulation Solution to Accelerate Chip Development

FPGA Prototyping in Practice: Addressing Peripheral Connectivity Challenges

S2C Advances RISC-V Ecosystem, Accelerating Innovation at 2025 Summit China

Double SoC prototyping performance with S2C’s VP1902-based S8-100