Bragadeesh Suresh Babu turned down an offer from Google’s Tensor Processing Unit team to build something bigger. The IIT Madras alumnus and former Fractile chip engineer just raised $1.7 million in pre-seed funding for Tattvam AI, a London-based deeptech startup building AI systems that automate semiconductor chip design.
The round was led by Seedcamp, with participation from EWOR, Entropy Industrial Ventures, Concept Ventures, and legendary semiconductor angel Stan Boland (former founder/CEO at Icera, acquired by NVIDIA).
The pitch? What currently takes 2-3 years of manual engineering work to design a custom chip — Tattvam AI aims to compress it into weeks using AI that “understands circuits from first principles.”
This isn’t about making existing chip design tools faster. It’s about replacing human designers entirely for specific tasks.
Table of Contents
The Problem: Custom Silicon Is Critical, But Designing It Takes Years
The world is racing toward custom silicon. Unlike general-purpose chips (CPUs, GPUs) built to handle everything, custom silicon refers to specialized processors optimized for specific workloads — AI training, AI inference, autonomous vehicles, drug discovery, cryptography.
The performance gains are staggering: custom chips can deliver up to 100x performance improvements over general-purpose hardware while consuming significantly less power. That’s why:
- Google built Tensor Processing Units (TPUs) for AI workloads
- NVIDIA partnered with Groq on specialized AI inference chips
- UK startups Fractile and Olix are building custom processors
- Cerebras built wafer-scale AI chips
- Tenstorrent (Jim Keller’s company) is building RISC-V AI accelerators
The stakes are enormous. As AI models balloon to trillions of parameters and applications demand massive computational power, custom silicon is becoming a critical competitive advantage.
But there’s a bottleneck: chip design still takes 2-3 years of painstaking manual work, and the pool of engineers who can do it is vanishingly small.
Why AI Tools Haven’t Solved This Yet
You’d think AI would’ve automated chip design by now. After all, AI is already writing complex code at record speed (GitHub Copilot, Cursor, Replit). But chip design is harder.
Current AI tools — even advanced LLMs — struggle with chip design because it demands deep structural understanding:
| Software Engineering | Chip Design |
|---|---|
| Logical reasoning (if-then-else) | Physical constraints (power, heat, timing) |
| Debugging through iteration | Fabrication errors cost millions |
| Single-layer abstraction | Multi-layer interdependencies (logic, physical layout, power delivery) |
| Forgiving error margins | Zero tolerance for timing violations |
Bragadeesh frames it perfectly: “Chip design is fundamentally a reasoning problem over an enormous search space, not unlike the kind of reasoning needed to solve hard problems in mathematics. Current AI tools struggle with the deep structural understanding that chip design demands.”
Tattvam AI’s Approach: A Reasoning Model That Understands Circuits
Tattvam AI isn’t building a better EDA (Electronic Design Automation) tool. It’s building a reasoning model that understands:
- Circuit constraints (power, area, timing)
- Trade-offs (speed vs power consumption)
- Interdependencies (changing one circuit block affects ten others)
The system is trained to think like a world-class chip design engineer — understanding circuits from first principles and autonomously solving complex design tasks.
If successful, this compresses the 2-3 year design cycle for custom silicon down to weeks, making custom chips accessible to far more companies, enabling rapid iteration, and dramatically reducing development costs.
The Founding Team: IIT Madras → CoMind → Fractile → Tattvam AI
Bragadeesh Suresh Babu (CEO) isn’t a first-time founder with zero domain expertise. He’s a mathematician-turned-chip-engineer with deep credentials:
- IIT Madras alumnus (India’s #1 engineering institution, alma mater of Perplexity CEO Aravind Srinivas)
- Early engineer at CoMind (UK-based brain-monitoring startup that recently raised $100 million)
- One of the earliest engineers at Fractile (UK chip startup building custom AI processors)
- Competitive mathematics background (olympiad-level problem-solving skills)
He co-founded Tattvam AI with Lannan Jiang, who’s been developing chips at a research lab at ETH Zurich — one of Europe’s top technical universities.
The combination of mathematical rigor + hands-on chip design experience + AI expertise is exactly what this problem demands. Stan Boland, whose companies Icera and Element 14 were acquired by NVIDIA and Broadcom respectively, is backing them:
“Bragadeesh is one of the most driven, energetic and compelling young founders in today’s chip industry. His conviction that Tattvam AI will dramatically speed up the complex and iterative process of using EDA tools and models to design chips, cutting timelines from years to weeks, is sure to be embraced by the world’s top teams.”
The IIT Madras Semiconductor Legacy
Bragadeesh’s IIT Madras pedigree isn’t incidental — the institution is India’s epicenter for semiconductor innovation:
SHAKTI Microprocessor Project: Led by Prof. V. Kamakoti, IIT Madras developed India’s first indigenous industrial-grade RISC-V processor. The project has already produced three chips fabricated at SCL Chandigarh — RIMO (2018), MOUSHIK (2020), and IRIS (2025, in collaboration with ISRO for space applications).
Bharat Semiconductor Research Centre: Established at IIT Madras to build India’s design and research capabilities with $2 billion government backing.
Mindgrove Technologies: IIT Madras-incubated startup launching the V2600 SoC for edge AI surveillance by late 2026.
InCore Semiconductors: Co-founded by Arjun Menon (ex-SHAKTI project), building RISC-V processors commercially.
IIT Madras has become India’s de facto semiconductor talent factory — and Tattvam AI is its latest global export.
The Market Opportunity: Billions in Custom Silicon Design
The total addressable market is massive:
- EDA software market: $15+ billion annually (Synopsys, Cadence, Siemens dominate)
- Custom chip design services: $10+ billion annually (ARM, Qualcomm, Broadcom, TSMC design teams)
- Fabless semiconductor companies: $200+ billion market cap collectively (NVIDIA, AMD, Qualcomm, Broadcom, MediaTek)
If Tattvam AI can automate even 20% of the chip design process, the value creation is staggering. Companies could:
- Design custom chips for niche applications previously too expensive to justify
- Iterate on designs 10x faster, enabling rapid product evolution
- Reduce dependence on scarce chip design talent (there are ~50,000 chip designers globally vs. ~30 million software engineers)
What the $1.7M Pre-Seed Will Fund
Tattvam AI plans to launch its first product in the coming months as it works with partners to accelerate next-generation chip development. The funding will go toward:
- Product development — completing the AI reasoning model and design automation pipeline
- Early customer pilots — working with chip startups, fabless companies, and design houses to validate the platform
- Team expansion — hiring AI researchers, chip design engineers, and GTM talent
The fact that Seedcamp (Europe’s top seed-stage fund) led the round, and that Stan Boland (NVIDIA/Broadcom exits) invested personally, signals serious conviction from people who’ve built and exited billion-dollar chip companies.
The Competition: EDA Giants vs. AI-Native Upstarts
Tattvam AI is entering a market dominated by three EDA giants:
- Synopsys ($85B market cap) — dominant in chip design automation
- Cadence ($85B market cap) — leader in verification and physical design
- Siemens EDA (formerly Mentor Graphics) — strong in automotive and industrial chips
But these incumbents are software companies trying to bolt AI onto decades-old tools. Tattvam AI is AI-native from day one — the entire system is designed around AI reasoning, not traditional rule-based automation.
Other AI-native chip design startups emerging include:
- Intrinsic ID (Netherlands) — AI-powered chip security
- Etched (US) — custom AI chips for transformers
- Various stealth-mode YC/a16z-backed startups
The race is on to see who can crack AI-automated chip design first.
The India Angle: From SHAKTI to Tattvam AI
Tattvam AI is technically a London-based company, but its roots are deeply Indian. Bragadeesh’s IIT Madras background, India’s booming semiconductor ecosystem, and government initiatives like the India Semiconductor Mission and Digital India RISC-V (DIR-V) create a powerful tailwind.
For Indian deeptech startups and chip design talent, Tattvam AI’s playbook is instructive:
- Build deep domain expertise (IIT Madras SHAKTI project, Fractile, CoMind)
- Solve hard technical problems (AI for chip design, not another SaaS wrapper)
- Target global markets (London HQ, Seedcamp funding, NVIDIA/Broadcom angels)
- Move fast (first product launching months after fundraise)
The Verdict: High-Risk, Higher-Reward
Automating chip design with AI is extraordinarily hard. If it were easy, Synopsys and Cadence would’ve done it already. The technical challenges are immense, the competition is fierce, and the validation timeline is long (chips take months to fabricate and test).
But if Tattvam AI succeeds, they won’t just build a valuable company — they’ll democratize custom silicon and unlock an entirely new category of computational power.
Bragadeesh turned down Google TPU. Stan Boland bet his reputation on backing him. Seedcamp wrote a check.
That’s not hype. That’s conviction.
Follow the latest Indian deeptech startups, semiconductor industry news, and AI innovation at Technosports.
