TYLsemi has closed an oversubscribed early-stage funding round, securing $43 million to advance its AI chiplet platform development.
This substantial investment aims to disrupt traditional semiconductor manufacturing, bringing advanced AI capabilities to a wider market. TYLsemi states its approach can reduce the costs of custom AI silicon development by almost 50% and complete it in less than half the time, addressing a critical need in the rapidly evolving AI sector.
Funding fuels custom AI silicon development
The $43 million early-stage funding round positions TYLsemi to advance its core mission: transforming custom AI accelerator design. This efficiency gain is crucial in a market projected to reach a colossal $604 billion by 2033, where custom silicon XPUs, or application-specific integrated circuits, for hyperscaler workloads represent the fastest-growing segment.
Mohit Gupta, TYLsemi’s Founder and CEO, articulated the market’s demand, stating, “The AI accelerator market is on track to reach $604 billion by 2033 and custom silicon XPUs built for specific hyperscaler workloads are the fastest-growing segment. 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.”
Gupta added that TYLsemi aims to close this gap with standards-based chiplets, combined with UCIe-based die-to-die connectivity, XPU-aware design, packaging, and integration. This offers customers a fast, proven path to AI-era silicon. The company believes this approach simplifies chip development, making it faster, less risky, and more accessible across the AI ecosystem.
Experienced leadership drives chiplet innovation
TYLsemi’s co-founders, Mohit Gupta and Sunil Bhardwaj, bring extensive experience from senior engineering and executive roles at prominent semiconductor firms. Both previously served as executives at Alphawave IP Group Plc, a British startup that Qualcomm Inc. acquired for approximately $2.4 billion last summer.
Their professional backgrounds also include leadership positions at SiFive Inc., Cadence Design Systems Inc., and Rambus. This pedigree offers TYLsemi a deep understanding of the challenges and opportunities within the chip design landscape, attracting investors like Matter Venture Partners.
Chiplet architecture redefines AI accelerators
The semiconductor industry is shifting towards chiplet-based architectures. This approach integrates multiple specialised smaller chips onto a single package, moving away from traditional monolithic designs where all functions reside on one die. Modularity is particularly advantageous for AI, where diverse computational requirements demand highly customised and scalable solutions.
Wen Hsieh, a Partner at Matter Venture Partners, highlighted TYLsemi’s disruptive potential. He noted that the company’s ability to speed up advanced silicon development and make it more cost-efficient will provide a defining advantage to AI companies. This advantage directly translates into quicker product cycles and reduced capital expenditure for businesses building AI infrastructure.
Standards-based design for broader access
TYLsemi capitalises on emerging standards like Universal Chiplet Interconnect Express (UCIe) for die-to-die links. This commitment to open standards ensures compatibility and interoperability across different chiplet manufacturers. It fosters a more dynamic and competitive ecosystem, allowing vendors to focus on optimising specific functions.
Zvika Orron, Managing Partner at Viola Ventures, shared this sentiment, stating, “It has built the chiplet platform that the entire AI silicon industry desperately needs.” He added that TYLsemi can serve both the scale of hyperscalers and the speed requirements of smaller, emerging AI companies.
This broader accessibility could foster a new wave of innovation across the AI ecosystem. Shri Dodani, Founder and Managing Partner at GHOVC, further noted that TYLsemi’s standards-based chiplets enable much lower-cost chip development for both startups and established silicon and system developers, allowing parallel development of multiple chip generations.
TYLsemi’s product portfolio for custom AI
TYLsemi’s offerings form a comprehensive suite designed to simplify and accelerate the creation of custom AI silicon. Its product roadmap addresses critical components for high-performance AI systems, covering data transfer, power delivery, memory management, and full-stack integration. These standardised, production-ready chiplets are built for multi-die AI systems.
TYL.IO: high-bandwidth connectivity
At the core of TYLsemi’s portfolio is TYL.IO, a family of input/output connectivity chiplets. These are engineered for high-bandwidth communication, essential for the immense data flows demanded by modern AI applications. TYL.IO supports established protocols such as PCIe, ESUN, and UALink.
The company also plans for future integration of co-packaged silicon photonics, promising even greater data throughput. These connectivity solutions are crucial for linking different chiplets effectively and efficiently within an AI system, preventing bottlenecks and ensuring optimal performance.
Efficient power and memory solutions
Complementing TYL.IO are TYL.Power and TYL.Mem. TYL.Power is a packaged power delivery system that incorporates an integrated voltage regulator chiplet, designed to maximise energy efficiency for XPUs. Efficient power delivery is paramount in AI accelerators, where high computational loads can lead to significant energy consumption.
TYL.Mem, currently under development, will provide versatile memory connectivity chiplets. Memory bandwidth and latency are critical factors in AI performance, especially for large language models and complex neural networks. Specialised memory chiplets allow designers to tailor memory configurations to their AI workload requirements, enhancing overall system efficiency.
TYL.Forge for comprehensive integration
TYL.Forge represents an end-to-end platform for integrating custom compute and fabric chiplets. This full-stack offering includes intellectual property, foundry management, advanced packaging, and production-readiness. TYL.Forge simplifies the entire design process, from architectural definition to final deployment.
This comprehensive platform allows customers to define their own XPU, compute, and fabric designs. It leverages TYLsemi’s expertise in managing the complex supply chain, addressing a significant pain point for companies looking to develop highly customised AI silicon without extensive in-house semiconductor design teams, thereby accelerating time to market and reducing risk.
This focus on making development less risky aligns with broader discussions around AI infrastructure faults and security concerns.
Strategic partnerships and market availability
TYLsemi has forged a crucial partnership with Taiwan Semiconductor Manufacturing Co. (TSMC), one of the world’s leading semiconductor foundries. This collaboration means TYLsemi will provide samples of its TYL.IO and TYL.Power chiplets to qualified customers through TSMC in 2027.
The availability of samples next year marks a significant milestone, allowing potential clients to evaluate the performance and integration capabilities of TYLsemi’s offerings. For its comprehensive TYL.Forge platform, the company is already actively engaging lead customers, demonstrating confidence in its broader, full-stack solution.
This structured rollout indicates a clear path to market adoption and scalability, positioning TYLsemi to become a key player in the evolving landscape of AI silicon development. The company’s strategy centres on using standards-based, production-ready chiplets combined with custom design, packaging, and supply-chain ownership.
