TYLSemi raises $43 million to build chiplets that democratize custom AI chip design

TYLSemi Secures Funding to Challenge Proprietary AI Chip Market

TYLSemi raises $43 million in early-stage funding to transform how companies design custom AI chips2

2

. The startup, founded by Mohit Gupta and Sunil Bhardwaj—executives from recent Qualcomm acquisition AlphaWave—launched less than a year after that deal closed. Matter Venture Partners led the funding round, with participation from Viola Ventures, GHOVC, and Egis Technology, plus strategic investors from the semiconductor and AI infrastructure ecosystem3

3

.

Breaking Free from Proprietary Lock-In with Open Standards

The custom AI chip design market is booming, with major companies like Meta Platforms working with Broadcom and others to develop custom semiconductors. Both Broadcom and rival Marvell Technology control proprietary technology for making chips communicate at high speed, forcing companies to work exclusively with them to access these capabilities. TYLSemi takes a different approach by supplying building blocks of custom AI chips called chiplets based on open industry standards, allowing customers to mix and match technologies from multiple providers3

3

. "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".

Chiplet-Based Design Cuts Costs and Development Time

TYLSemi aims to make custom AI accelerators more accessible by addressing the immense costs and complexity that have traditionally limited development to tech giants[2](https://siliconangle.com/2026/07/14/custom-ai-chip-design-startup-tylsemi-l

aunches-43m-early-stage-funding/). Companies like Google with its tensor processing units and Amazon Web Services with Trainium chips have invested hundreds of millions of dollars in custom application-specific integrated circuits known as XPUs2

2

. By combining standards-based and production-ready chiplets with custom design, packaging, and supply-chain ownership, TYLSemi believes it can reduce custom AI silicon development costs by almost 50% and complete projects in less than half the time2

2

.

Modular Components Address Physical Limitations of AI Hardware

Chiplets serve as smaller, specialized modular components that perform distinct functions like processing, memory, or power management2

2

. These can be integrated into a single 3D package that acts as a unified, high-performance system, bypassing the physical limitations of traditional monolithic chips printed on a single silicon wafer2

2

. The company has already secured engagements with tier-one customers and offers production-ready chiplets including TYL.IO for input/output connectivity supporting PCIe, ESUN, and UALink protocols, with future roadmaps calling for silicon photonics integration2

2

.

Market Forecast Points to $604 Billion Opportunity

Co-founder and CEO Mohit Gupta noted that the market for AI accelerators is forecast to hit $604 billion annually by 2033, with custom silicon XPUs expected to be the fastest-growing segment2

2

. "At that scale, chiplet-based design is no longer optional, yet there is no pure-play chiplet company serving this market with a full portfolio," he explained2

2

. Recent advances in 2.5D and 3D packaging technologies now make it possible to integrate heterogeneous dies and dynamic random-access memory into a single system, while standardized interconnects like UCIe make chiplet design practical and scalable for a broader range of companies2

2

.