Itzik Gilboa

Somewhere in the gap between a Haifa classroom and a Silicon Valley boardroom, Itzik Gilboa figured out that the semiconductor industry had a problem nobody wanted to admit: its factories were running on instinct, spreadsheets, and expensive tribal knowledge. Not because the engineers weren't brilliant. Because the complexity of a modern fab - thousands of process steps, millions of wafers, variables cascading in every direction - simply exceeds human bandwidth. He decided to fix that.

Gilboa started at the Technion - Israel Institute of Technology, that storied hill in Haifa where Israel's engineering culture took root. He came in studying aerospace and left with a second degree in materials science. The pivot tells you something: he was always drawn to materials that do extraordinary things under pressure. Chips, it turns out, qualify.

From Technion he moved through the industry's foundational companies. Cypress Semiconductor made him a Director of Wafer Fab - a role that forces you to understand why fabs run slow, run hot, and occasionally don't run at all. Then SanDisk, where as Sr. Director of Strategic Programs he steered product cost management and ran the SanDisk/Toshiba Alliance joint ventures - one of the most intricate partnerships in chip history. Then Western Digital, heading Mobile & Components Business. By the time he arrived at minds.ai, he had spent two decades accumulating precisely the vocabulary a fab refuses to share with outsiders.

minds.ai was founded in 2014 out of a simple, audacious premise: multi-agent reinforcement learning - the same class of algorithms that mastered Go and chess - could be put to work scheduling semiconductor fabs in real time. Not simulating. Not advising. Actually running. The company's Maestro platform deploys AI agents across fab operations, optimizing cycle time, on-time delivery, and utilization with a continuous learning loop that gets smarter the longer it runs.

That's the technical argument. The strategic one is starker. As chip demand accelerates - driven by AI, EVs, and every connected device humans now carry - fab throughput has become a geopolitical variable. A day of lost yield at a leading fab isn't a line item. It's a supply chain event. Gilboa understood this years before the chip shortage of 2021 made it front-page news.

Gilboa doesn't just run his own company. He advises three of the most significant deep-tech startup communities in Silicon Valley: Alchemist Accelerator, which focuses on enterprise startups; Silicon Catalyst, the world's only incubator exclusively focused on semiconductors; and Berkeley SkyDeck, UC Berkeley's flagship accelerator. He also advises Algorized and holds an executive membership in the International LEAP Network.

That's not resume decoration. That's a man who spent 20 years absorbing how the chip industry actually works - its procurement cycles, its qualification processes, its tolerance for risk, its resistance to change - and now passes that knowledge to founders who are trying to change it. When a 28-year-old founder is pitching an AI solution to a fab engineer who has spent 30 years tuning a process manually, Gilboa knows exactly how that conversation goes. He's been both people.

His MBA from San Jose State anchored the business instinct. His Stanford program in Strategic Decision and Risk Management sharpened the judgment under uncertainty. But the real education happened at the SanDisk/Toshiba Alliance - managing a joint venture between two companies with different cultures, different incentives, and billions of dollars in shared infrastructure. That's the kind of experience that either breaks you or makes you extraordinarily good at navigating complexity.

The AI conversation in 2026 is dominated by models and tokens. Itzik Gilboa is working on something harder: getting AI to operate reliably inside the most unforgiving physical environments on earth. A semiconductor fab runs 24/7/365. Decisions made in milliseconds affect wafers worth thousands of dollars each. There is no "undo." The AI systems that thrive in this environment aren't the ones with the best benchmarks on abstract tasks - they're the ones trained on the specific, messy, real-world data of production fab operations.

That's what minds.ai has been building since 2014. Not a product for the demo room. A product for the fab floor. Gilboa came from that floor, respects it, and has spent the last decade making sure his team does too.

His aspiration is straightforward and vast: make semiconductor manufacturing as intelligent and adaptive as the chips it produces. In a world where chip supply is a strategic resource and every percentage point of yield improvement ripples into billions of dollars of economic impact, that might be the most consequential AI application nobody's talking about at a keynote.