The semiconductor market will hit $1 trillion in annual revenue by 2030, according to McKinsey and Company. Anna Goldie and Azalia Mirhoseini already proved that AI can design chip layouts better than human engineers. Now they’ve launched Ricursive Intelligence. At TechCrunch Disrupt 2026, they showed where this goes next. It’s bigger than most investors realize.
Why This Is Happening Right Now
Goldie and Mirhoseini aren’t newcomers to this problem. In 2021, they published research in Nature showing that a reinforcement learning model could place components on a chip faster and more efficiently than senior human designers at Google. The paper landed hard in the chip design world. According to Nature, the AI completed chip placements in under six hours that would take trained engineers several months.
The semiconductor shortage of 2021 through 2023 cost the global economy an estimated $500 billion, according to AlixPartners. The industry’s response was to pour money into new fabrication plants. But fab capacity only solves part of the problem. Design speed is the other bottleneck. Slower design means slower time to market. That means lost revenue and lost ground to competitors. Ricursive Intelligence is betting that compressing the design timeline is worth more than building more factories.
The $12 Billion Market Most People Are Ignoring
Most people will read the TechCrunch Disrupt 2026 coverage and think this is an interesting research story. Maybe useful in a few years. That’s the mindset that keeps people on the sidelines while capital moves.
The rich mindset asks a different question. Who controls the design tools?
Right now, two companies dominate electronic design automation (EDA). Synopsys and Cadence hold over 60 percent of the global EDA software market combined, according to IDC. Their tools cost millions of dollars per seat each year. Every major chipmaker pays them. Nvidia pays them. Apple pays them. Samsung pays them. The EDA market is worth roughly $12 billion per year, according to EE Times.
At TechCrunch Disrupt 2026, Goldie and Mirhoseini showed something that changes the math on all of that. Their system at Ricursive Intelligence doesn’t just assist with chip design. It runs the design loop on its own. The model generates layout proposals, simulates performance, finds failures, and revises the design without human input at each step. According to the founders, their system reduces chip design cycles from 18 months to under 90 days for certain chip architectures.
If that holds at scale, it doesn’t just speed up the industry. It threatens to replace the most expensive software in the industry. Synopsys alone posted $6.1 billion in revenue in fiscal 2025, according to their annual report. Cadence posted $4.6 billion over the same period, according to Cadence’s investor relations filings. The people who build AI native design tools that undercut those prices will collect a lot of those dollars.
I think the real play is the IP that the AI generates. Every layout the model produces is a potential patent. The company that trains the model owns the output. That’s a completely different asset class from traditional software. It’s not subscription revenue. It’s ownership of the designs that future chips will be built on.
The founders were careful at Disrupt 2026. Their system works best on certain chip types, not the most complex CPUs or GPUs yet. But Goldie made a point worth remembering. The model improves more from each design iteration than a human engineer does. Compound improvement over thousands of design cycles is a serious advantage.
What This Means for You
If you invest, watch the EDA market closely. Any company that can shorten a chip design cycle from 18 months to 90 days will attract serious enterprise contracts. Taking 10 percent of Synopsys’s customer base would be a billion dollar business. That’s not a speculative number. It’s basic market math.
If you build hardware products, this changes your timeline math. Shorter design cycles mean faster iteration. A startup building custom silicon for a niche application no longer needs a three year runway just for the design phase. The barrier to custom hardware is falling, and that means more competition at every level of the hardware stack.
If you’re thinking about starting a company in this space, the window is open right now. AI hardware and chip design tools are early enough that a small team with the right background can compete. If you’re setting up your entity, Inc Authority offers free LLC formation and can get you legally structured without the legal fees that drain runway before you ship a single product.
One practical note on moving fast in any deal: hardware licensing agreements and joint development contracts crawl if your paperwork does. I use signNow for all my contract workflows. Documents go from draft to signed in hours instead of weeks. When your competitive window is 90 days instead of 18 months, slow signatures cost you real money.
The Bottom Line
AI designing its own hardware isn’t a future story. It’s a 2026 story. Goldie and Mirhoseini showed real results at TechCrunch Disrupt 2026. The EDA market sits at $12 billion per year. The chip design bottleneck costs the global economy hundreds of billions when it breaks. The people betting that Synopsys and Cadence hold their positions forever are going to lose capital. The people who understand that AI is eating hardware design tools next are going to be very well positioned.
Frequently Asked Questions
What is Ricursive Intelligence and what does it do?
Ricursive Intelligence is a startup founded by AI researchers Anna Goldie and Azalia Mirhoseini. The company builds AI systems that design chip layouts faster and more efficiently than human engineers. Their system targets reducing chip design cycles from 18 months to under 90 days for certain chip architectures, according to the founders at TechCrunch Disrupt 2026.
Who are Anna Goldie and Azalia Mirhoseini?
Both researchers published a landmark paper in Nature in 2021 showing that reinforcement learning could outperform human engineers at chip placement inside Google. Goldie focuses on algorithmic approaches to chip layout optimization. Mirhoseini applies machine learning to engineering and scientific problems. They founded Ricursive Intelligence to commercialize this research.
How does AI chip design work at Ricursive Intelligence?
The AI generates possible chip layouts, simulates performance, identifies failure points, and revises the design automatically. This loop runs thousands of times per hour. According to the founders, the model improves faster with each design cycle than a human engineer would, which means the system gets more capable the more designs it processes.
What is the EDA market and why does it matter for AI chip design?
Electronic design automation is the software category chip designers use to plan, simulate, and verify chip layouts. According to EE Times, the EDA market is worth roughly $12 billion per year. Synopsys and Cadence control over 60 percent of it, according to IDC. AI native chip design tools that compete on speed and price represent a direct threat to that dominance.
Should founders and investors pay attention to AI hardware design right now?
Yes. The design cycle compression that Ricursive Intelligence demonstrated at TechCrunch Disrupt 2026 is commercially significant today, not in five years. Investors should watch EDA incumbents for margin pressure. Founders in adjacent hardware markets should rethink their timeline assumptions because the barrier to custom silicon is falling faster than most roadmaps account for.
