The semiconductor industry has a problem that's as old as it is expensive: designing advanced computer chips takes too long and costs too much. Cognichip, a startup founded in 2024, believes artificial intelligence can fix that — and it just secured $60 million in fresh funding to prove the idea works.
The Bottleneck in Silicon
Modern chip design is one of the most complex engineering feats on the planet. A cutting-edge GPU like Nvidia's Blackwell architecture packs roughly 104 billion transistors onto a single piece of silicon. Getting from an initial concept to mass production typically takes three to five years, with the design phase alone consuming up to two years before physical fabrication even begins. That timeline carries enormous risk. Markets shift, customer needs evolve, and competitors don't sit still. By the time a chip reaches production, the opportunity it was built for may have already passed.
Cognichip's founder and CEO, Faraj Aalaei, sees this as a solvable problem. His company is building a deep learning model purpose-built for chip design — one that works alongside human engineers rather than replacing them. The goal is to bring the same kind of AI-assisted acceleration that software developers have enjoyed with coding copilots into the world of hardware design. Aalaei has said the technology could slash chip development costs by more than 75 percent and cut timelines by more than half.
Big-Name Backers Join the Board
The $60 million round was led by Seligman Ventures, with a particularly notable participant: Intel CEO Lip-Bu Tan, who invested through his venture firm Walden Catalyst Ventures and will take a seat on Cognichip's board. Umesh Padval, a managing partner at Seligman, will also join the board. Including earlier rounds, Cognichip has now raised $93 million in total since its founding.
Padval has framed the current moment as one of historic opportunity, calling the flood of capital into AI infrastructure the largest investment wave he has witnessed in four decades. In his view, a boom in semiconductors and hardware naturally creates a boom for companies like Cognichip that serve the chip design ecosystem.
Training an AI on a Secretive Industry
One of Cognichip's central challenges — and what it positions as its core advantage — is data. Unlike the software world, where vast open-source repositories provide ample training material for AI coding assistants, the semiconductor industry is notoriously secretive. Chip designers fiercely protect their intellectual property, and there is no equivalent of GitHub for hardware design files.
To overcome this, Cognichip has built its own proprietary and synthetic datasets and has licensed data from industry partners. It has also developed secure procedures that allow chipmakers to train Cognichip's models on their own confidential data without exposing it to outsiders. The company chose to train a domain-specific model from the ground up rather than fine-tuning a general-purpose large language model, arguing this yields better results for such a specialized task.
Where proprietary data isn't available, Cognichip has leaned on open-source alternatives. During a hackathon at San Jose State University, electrical engineering students used the model to design CPUs based on the open-source RISC-V architecture, demonstrating that the tool can be effective even without access to closely guarded commercial designs.
A Crowded and Well-Funded Arena
Cognichip is not the only startup chasing AI-powered chip design. ChipAgentsAI closed a $74 million extended Series A in February, while Ricursive raised a massive $300 million Series A in January at a reported $4 billion valuation. Established electronic design automation giants like Synopsys and Cadence Design Systems also loom large, with decades of industry relationships and deeply entrenched tools.
The competition is fierce, but so is the prize. As AI models grow larger and more power-hungry, the demand for new, specialized chips is accelerating faster than the industry's traditional design processes can keep up. Any company that can meaningfully compress the chip design cycle stands to capture an outsized share of a rapidly expanding market.
What Comes Next
For now, Cognichip cannot yet point to a finished chip designed with its platform, and it has not publicly named any of the customers it says it has been collaborating with since September. The real test will come when a chip designed with significant AI assistance reaches fabrication — and performs. Until then, the $93 million bet is that AI can do for hardware what it has already begun doing for software: make humans faster, cheaper, and more creative.
