A $60 million bet on a simple but radical idea: what if the thing holding AI back is that humans are still designing the hardware by hand?
On April 1, 2026, Cognichip announced a $60 million Series A funding round — oversubscribed, led by Seligman Ventures, and backed by one of the most powerful names in the semiconductor world: Lip-Bu Tan, the current CEO of Intel, who joined the company's board. With this round, Cognichip's total funding reaches $93 million since its founding in 2024.
The announcement lands at a moment when the entire AI industry is confronting a hard ceiling. Models are getting smarter. Data centers are expanding. But the chips that run everything — the physical silicon that makes AI possible — are struggling to keep pace.
The Problem No One Talks About Enough
Designing an advanced chip today takes three to five years from conception to mass production. The design phase alone can stretch two years before physical layout even begins. The latest Nvidia Blackwell GPU contains 104 billion transistors. Each one needs to be positioned, verified, and optimized across digital, analog, and mixed-signal domains simultaneously — a process that costs hundreds of millions of dollars per chip family.
Here is the circular trap the industry is stuck in: AI capabilities are advancing faster than the hardware that runs them. But building better hardware requires time and resources that scale poorly. Cognichip's founder and CEO, Faraj Aalaei, calls this a "circular constraint" — AI is limited by hardware, but hardware cannot evolve fast enough to keep up with AI.
The result? Every powerful new model you use today is already running on chips that were designed years ago. The hardware is always behind.
What Cognichip Actually Built
Cognichip is not building another electronic design automation (EDA) tool — the category dominated by incumbents Synopsys and Cadence Design Systems. Instead, it created what it calls ACI — Artificial Chip Intelligence — a full-stack, physics-informed foundation model trained specifically on chip design data.
The distinction matters. General-purpose AI models like GPT or Claude were trained on text and code. They can help engineers write scripts or documentation, but they have no native understanding of heat dissipation, power distribution, transistor physics, or the complex interdependencies between a chip's digital and analog layers.
Cognichip's ACI was built from the ground up to understand these constraints. It fuses logic-based and physics-based reasoning to navigate design spaces that no general-purpose AI can reach. Aalaei describes it not as a tool, but as an engineering collaborator: tell it what result you want, and it produces the design.
The Numbers That Make This Real
The potential impact is measurable and specific. According to Cognichip:
Chip development costs can be reduced by more than 75%. Design timelines can be cut by more than half, from three to five years down to potentially 18 months or less. The platform reduces engineering effort by up to 50% on individual design tasks. Cognichip is already working with more than 30 semiconductor design companies, including several of the industry's largest.
For context: semiconductor design is a market where a single chip family can cost $500 million to $1 billion to develop. A 75% cost reduction would be one of the most consequential efficiency gains in the history of the industry.
The Virtuous Cycle — And Why It Changes Everything
Here is the idea that makes Cognichip more than just a startup story. It is about what happens when the loop closes.
Today, AI makes chips possible. Tomorrow, AI will design the chips that make the next generation of AI possible. Faster, more efficient chips lower the cost of training and running AI models. Cheaper AI unlocks more research cycles. More research produces better AI. Better AI designs even more efficient chips.
This is the virtuous cycle — and it has no obvious ceiling.
Umesh Padval, Managing Partner at Seligman Ventures and a 40-year semiconductor veteran, described the current moment as the largest wave of capital into AI infrastructure he has ever seen. Intel CEO Lip-Bu Tan, who previously led Cadence Design Systems through its transformation into a global EDA leader, called the semiconductor industry at "a critical juncture" where an AI-first framework will unlock massive global opportunity.
The Competition Is Already Heating Up
Cognichip is not alone. ChipAgents closed a $74 million extended Series A in February 2026. Ricursive raised a $300 million Series A in January 2026. Synopsys and Cadence are embedding AI into their own workflows. The race to automate chip design has begun in earnest.
What differentiates Cognichip is its data strategy. Unlike software development, where billions of lines of open-source code are freely available to train AI models, chip design data is tightly guarded intellectual property. Cognichip built its own proprietary dataset, supplemented with synthetic data and licensed information from partners, and developed secure training pipelines that allow chipmakers to train the model on their own IP without exposing it externally.
In a hackathon at San Jose State University, engineering students used Cognichip's platform to design functional CPUs based on the open-source RISC-V architecture — a meaningful early proof of concept.
What the $60 Million Will Do
Cognichip plans to expand operations, deepen partnerships with semiconductor firms, and build global presence. The oversubscribed nature of the round — meaning investors wanted to put in more money than Cognichip needed — signals confidence that goes beyond typical early-stage bets.
The board now includes Intel's sitting CEO, one of the most experienced chip executives alive. This is not a speculative vote. It is a structural endorsement from inside the industry that AI-driven chip design is not a future possibility — it is the present direction.
The End of Scale Limits?
For a decade, the AI industry has scaled by adding more compute — more GPUs, larger data centers, bigger training runs. That approach is running into physical and economic walls. Power consumption, chip yields, and design costs are all becoming bottlenecks.
Cognichip's bet is that the solution is not to build more of the same chips faster. It is to design fundamentally better chips — chips optimized by AI for AI — in a fraction of the time and cost. If it works, the limits that have been slowing AI's hardware progress for years could begin to dissolve.
The virtuous cycle would be complete.
Want to follow every major AI move the moment it happens? Join thousands of readers at fabioraphaelli.com and get the biggest AI news every week.
Frequently Asked Questions
What is Cognichip and what does it do?
Cognichip is a semiconductor startup founded in 2024 that uses artificial intelligence to assist in designing computer chips. Its ACI (Artificial Chip Intelligence) platform uses physics-informed AI models to reduce chip development costs by over 75% and cut design timelines by more than half.
How much has Cognichip raised and who are its investors?
Cognichip raised $60 million in a Series A led by Seligman Ventures, with participation from SBI Investment, Mayfield, Lux Capital, and Intel CEO Lip-Bu Tan. Total funding reached $93 million since founding in 2024.
Why is AI-designed chip design important for the future of AI?
Current AI models are limited by the speed at which new chips can be designed and manufactured — a process that takes three to five years. If AI can design chips faster and cheaper, it creates a self-reinforcing cycle: better chips enable more powerful AI, which in turn designs even better chips.
How is Cognichip different from existing chip design tools?
Unlike general-purpose AI or traditional EDA tools from Synopsys and Cadence, Cognichip built a foundation model trained specifically on chip design data, with physics-based reasoning that understands heat, power, and transistor behavior — constraints that generic AI cannot handle.
What is the "virtuous cycle" in AI chip design?
This refers to the self-reinforcing loop where AI designs more efficient chips, which then enable more powerful AI, which in turn designs even better chips. This cycle has the potential to accelerate AI hardware development significantly.
Has Cognichip already designed a real chip?
As of April 2026, Cognichip has not publicly announced a commercially shipped chip designed with its platform. It is collaborating with more than 30 semiconductor firms and has demonstrated early capabilities through academic hackathons using open-source chip architectures.







