Databricks Architect Wins Top Computing Prize, Argues AGI Has Arrived

Matei Zaharia, the co-founder and chief technology officer of Databricks, recently learned he'd won the 2026 ACM Prize in Computing. He nearly overlooked the notification. "It was a surprise," Zaharia admitted.

His journey to this accolade began in 2009 with his PhD research at UC Berkeley. Under professor Ion Stoica, Zaharia built a framework to accelerate sluggish big data workloads. He released it as the open-source project Apache Spark. At a time when big data dominated tech conversations, Spark became a fundamental tool, propelling the young engineer to prominence.

Zaharia has since guided Databricks from a startup to a cloud data powerhouse, now central to many corporate AI strategies. The company has achieved a valuation of $134 billion and generates billions in revenue.

The Association for Computing Machinery recognized his cumulative impact with the award, which includes a $250,000 prize Zaharia plans to donate to charity.

Now an associate professor at Berkeley alongside his CTO role, Zaharia is focused on artificial intelligence. He holds a provocative view: "AGI is here already. It's just not in a form that we appreciate." He suggests the error is in judging AI by human benchmarks. A system that passes a bar exam by processing vast data isn't demonstrating human-like understanding; it's executing a different kind of task.

This anthropomorphism, he warns, creates risk. He points to AI agents designed as trusted assistants. "It's a security nightmare," Zaharia notes, citing potential for unauthorized access or spending. "Yeah, it's not a little human there."

His enthusiasm lies in AI's potential to automate complex research, from biology to data synthesis. He envisions a future where AI tools, free from fabrication, help people understand information rather than just build software. He calls this "AI for search, but specifically for research or engineering" - systems that diagnose mechanical problems, interpret non-visual data, or simulate molecular interactions. For Zaharia, the next breakthrough is leveraging what AI does uniquely well, not asking it to imitate us.

Source: TechCrunch