Venkat
Mattela

March 2020. The ink was barely dry.

Silicon Labs closed the acquisition of Redpine Signals' wireless connectivity business on March 12, 2020. The deal: $308 million. Mattela had spent a decade-plus building that company - the wireless IoT pioneer whose ultra-low-power solutions ran circles around Qualcomm and Broadcom by a factor of 26 in energy efficiency, and accumulated 130 foundational patents along the way.

April 2020. Ceremorphic was incorporated.

That's not a gap year story. That's a man who had already been thinking about the next problem for years. The R&D that became Ceremorphic's foundation started while Redpine Signals was still operating - a parallel track of curiosity that eventually consumed everything. When the wireless exit closed, Mattela walked straight into the thing he'd been building in the background.

The new problem: AI computing was scaling in the wrong direction. More power, more heat, more cost, more infrastructure. The industry's answer was always bigger. Mattela's answer was always more efficient - the same instinct that made Redpine Signals' chips 26x leaner than the competition is now being applied to the full stack of AI supercomputing.

Ceremorphic: Supercomputing on a Developer's Desk

The pitch for Ceremorphic sounds like hubris until you understand who is saying it. An AI supercomputing chip - the kind of thing that usually requires a server rack in a climate-controlled cage - small enough, efficient enough, and economical enough to sit next to a monitor. That's the stated goal. The QS1 chip is the first step.

What makes QS1 different is the process node and the architecture. Ceremorphic became one of the first companies - certainly the first startup at its scale - to tape out an AI supercomputing chip on TSMC's 5nm process. The QS1 pairs a 2GHz custom machine learning processor with a custom floating-point unit, a patented multi-threaded processing architecture called ThreadArch, video engines for metaverse workloads running at 1GHz, and PCIe 6.0/CXL 3.0 connectivity. It's designed for three hard requirements that most AI silicon treats as afterthoughts: reliability, security, and energy efficiency at scale.

When AI Meets Drug Discovery

In October 2023, Ceremorphic announced something that most chip companies would never attempt: a life sciences division. Not a partnership. Not a customer. An internal platform called BioCompDiscoverX - built on Ceremorphic's proprietary analog and AI technology - aimed at fundamentally changing how drugs are discovered.

The problem Mattela is targeting is specific: AI in drug discovery fails not because the algorithms are bad, but because the training data is wrong. Current approaches try to apply machine learning to incomplete or irrelevant biological datasets, producing models that look impressive in silico and fail in the clinic. Ceremorphic's answer is to generate better data through analog simulation - using circuits that can model biological systems more accurately than purely digital computation.

The BioCompDiscoverX platform represents, by the company's own accounting, more than 5,000 person-years of engineering effort across six-plus years. That math predates Ceremorphic's official 2020 founding, meaning Mattela was building this capability while still running Redpine Signals. The keynote he delivered at the AI Driven Drug Discovery Summit in Boston was the public debut of an effort that had been quietly accumulating for years.

Redpine Signals: The $308M Blueprint

Before Ceremorphic, there was Redpine Signals - a wireless semiconductor company Mattela built from the ground up into one of the IoT industry's most respected engineering shops. The company's ultra-low-power wireless solutions entered a market dominated by giants and proceeded to make them look wasteful. Redpine's chips achieved power efficiency up to 26 times better than competing solutions from the large incumbents.

That efficiency advantage wasn't an accident. It came from Mattela's core engineering philosophy - one that had been forming since his early career at Analog Devices, where he led VLSI development for networking silicon, and Infineon Technologies, where he directed micro-architecture on the TriCore MCU-DSP processor. Energy efficiency wasn't a feature at Redpine. It was the architecture.

By the time Silicon Labs came calling, Redpine Signals had accumulated 130 foundational wireless patents and built a development center in Hyderabad with approximately 200 engineers. Silicon Labs acquired the wireless connectivity assets for $308 million in March 2020 - a deal that validated not just the technology but the 10-year engineering bet Mattela had made on ultra-low-power design.

Forty Years, One Throughline

The Engineer Who Thinks in Decades

There is a consistency to Mattela's career that only becomes visible from far enough away. He starts at the edges of what's possible in power efficiency, builds until it's undeniable, and then exits to build the next version of the same bet at a different scale. At Analog Devices, it was networking silicon. At Infineon, it was processor architecture. At Redpine Signals, it was wireless IoT. At Ceremorphic, it's AI supercomputing.

Each move took the core insight - that efficiency is architecture, not optimization - and applied it to a harder problem. The jump from wireless IoT to AI supercomputing chips looks large from the outside. From Mattela's vantage point, it's the same equation solved for a bigger variable.

His India-first engineering model is another consistent thread. Both Redpine Signals and Ceremorphic put the majority of their engineering talent in Hyderabad - not as a cost play, but as a deliberate decision to build from a talent pool Mattela has understood and cultivated for decades. Roughly 90% of Ceremorphic's engineering work happens in India. The company is planning a $10 million annual investment in the Hyderabad center alone, with targets to reach 400 engineers in the city.

He said that in 2017. The consolidation he predicted happened exactly as he described - and Redpine Signals was one of the survivors, acquired at a premium. The quote is a small data point in a long pattern of being early and right about where the market goes.

The Record

• Founded and scaled Redpine Signals to 200+ engineers with 130+ wireless patents
• Sold Redpine Signals wireless division to Silicon Labs for $308 million (March 2020)
• First startup to tape out an AI supercomputing chip on TSMC 5nm process (Ceremorphic, 2022)
• Raised $50M Series A for Ceremorphic in January 2022
• Over 230 total US and international patents filed across career
• Wireless solutions at Redpine achieved 26x better energy efficiency vs. industry leaders
• Built Ceremorphic Life Sciences BioCompDiscoverX - 5,000+ person-year engineering effort
• Harvard Business School OPM IGNITE program graduate and invited speaker
• Keynote speaker at AI Driven Drug Discovery Summit, Boston
• India UK Achiever's Award 2025
• Presented at Linley Spring and Fall Processor Conferences (2022)

Links & Profiles