Semiconductor Startup Tylsemi Raises $43m in Seed Fund To Develop Modular Chip Tech

TYLSemi, a semiconductor startup founded by former AlphaWave executives shortly after Qualcomm acquired the company, has raised $43 million in seed funding to develop modular chip technology aimed at helping companies build custom artificial intelligence processors more quickly and at lower cost.

The fundraising comes as demand for custom AI chips accelerates beyond hyperscalers such as Meta, Microsoft, Amazon and Google, creating opportunities for startups seeking to lower the barriers to designing specialized semiconductors for AI workloads.

The company was founded by Mohit Gupta and Sunil Bhardwaj, who previously held senior positions at AlphaWave, the high-speed connectivity chip company acquired by Qualcomm. Their new venture aims to disrupt one of the AI semiconductor industry’s most lucrative businesses by replacing proprietary chip designs with standardized, interoperable building blocks.

The AI infrastructure boom has triggered a surge in demand for custom application-specific integrated circuits (ASICs), allowing cloud providers and large enterprises to reduce dependence on Nvidia’s increasingly expensive graphics processors.

Companies such as Meta Platforms have already partnered with Broadcom to develop proprietary AI chips, while Broadcom and Marvell Technology have emerged as dominant providers of the high-speed interconnect technologies that allow different parts of AI processors to communicate efficiently.

Those technologies have traditionally been available only through proprietary development partnerships, effectively locking customers into a particular vendor’s ecosystem and making it costly to switch suppliers.

TYLSemi is betting that the next phase of AI chip development will move away from that closed model. Instead of developing entire processors, the startup plans to supply chiplets—small semiconductor modules that perform specific functions and can be combined with components from multiple vendors using open industry standards.

Customers would then integrate these chiplets with other technologies before packaging them into complete AI processors tailored to their own workloads. The approach was introduced at a time in the semiconductor industry when chiplet architectures were increasingly replacing monolithic chip designs. By breaking complex processors into smaller modules, manufacturers can improve production yields, shorten development cycles, and reduce costs while allowing individual components to be upgraded independently.

“I feel progress happens with standardization,” Gupta told Reuters.

“Whenever you do proprietary lock-in, it’s a short-term game. Yes, you can squeeze (customers) given your position and whatnot, but it’s not healthy for the market.”

Expansion of AI infrastructure is creating openings for new players. Thus, the funding underscores growing investor confidence that the AI hardware market is expanding beyond established chipmakers to include companies building enabling technologies for custom silicon.

The seed round was led by Matter Venture Partners, with participation from Viola Ventures, GHOVC, and Egis Technology.

TYLSemi also disclosed that it secured strategic backing from unnamed companies across the global semiconductor and AI infrastructure ecosystem, suggesting established industry players see value in fostering an open alternative to proprietary AI chip platforms.

The startup enters the market as AI infrastructure spending continues to reach unprecedented levels. Hyperscalers are investing hundreds of billions of dollars annually in data centers and increasingly designing their own processors to improve performance while lowering operating costs. That trend has fueled explosive growth for Broadcom and Marvell, whose custom silicon businesses have become key beneficiaries of the industry’s shift toward AI-specific hardware.

However, their business models rely heavily on proprietary technologies that can limit customers’ flexibility.

TYLSemi is aiming to become a neutral supplier that enables companies to source different chip components from multiple vendors rather than depending on a single semiconductor partner. If widely adopted, that approach could lower development costs, shorten time-to-market and encourage greater competition across the custom AI chip ecosystem.

The company has emerged amid the growing wave of semiconductor startups being launched by veterans of established chip companies. As demand for AI infrastructure expands beyond the largest cloud providers, investors are now backing firms that specialize in critical parts of the semiconductor value chain rather than attempting to compete directly with industry leaders such as Nvidia.

While TYLSemi remains in its early stages, its focus on open chiplet standards aligns with a broader industry movement toward modular semiconductor design.