The CEO Who Believes AGI Is Already Here

Welcome back to In the Loop, TIME’s new twice-weekly newsletter about AI. If you’re reading this in your browser, why not subscribe to have the next one delivered straight to your inbox? Who to Know: Ali Ghodsi, CEO of Databricks The three most valuable private companies in the U.S. have big reputations: OpenAI, SpaceX, and Anthropic. But the fourth, Databricks, flies a little more under the radar. This company, which is currently raising funds at a valuation of $134 billion according to reports this week, is a quiet workhorse of the AI revolution. Databricks offers a platform where companies can combine unwieldy data from different sources, manage it easily, and train AI models with it. Its CEO, Ali Ghodsi, is a charismatic computer science professor who still teaches the occasional class at U.C. Berkeley, even as he runs a multibillion-dollar company. I spoke with him last week. On AGI — Artificial general intelligence has already been achieved, Ghodsi says. To justify this provocative statement, he casts his mind back to his experience working in a computer science lab two decades ago, discussing in vaunted terms with colleagues what AGI might look like: it would be able to talk, and reason, and spot patterns in huge quantities of data. “Now that we have it, it’s kind of like, never meet your heroes,” Ghodsi says with a laugh. “It’s just like, okay, that’s not that impressive.” Okay, but — Ghodsi has good reason to make this argument. Although OpenAI is one of his largest customers, Databricks’ bread and butter comes from the long tail of more normal companies that aren’t pursuing AGI, but simply want to use existing AI to do things with their data. In Ghodsi’s telling, Databricks is capitalizing on the size of this opportunity. “If all AI progress was frozen today, I think we have what we need to proceed with what we are doing,” Ghodsi says. “Not trying to build super-God… We’re much more focused on: how can we make AI useful today in organizations?” On AI agents —“Agents” was the buzzword of 2025 that never really delivered. By all means, AI improved radically this year, including by gaining new abilities to go and carry out web searches, or modify and run code. But big AI models still face reliability challenges that limit the length of time they can spend running autonomously, and we don’t have full-fledged AI coworkers yet. Ghodsi says Databricks is working on a solution to this problem: helping companies train smaller, laser-focused agents, using their own data. Those models end up being both cheaper to run and more reliable for specific tasks than frontier AI models like ChatGPT or Claude. “We can give you a good quality on the specific task, because we’re cheating and we’re doing boring AI, focusing just on specific tasks that you want help with,” Ghodsi says. If you have a minute, please take our quick survey to help us better understand who you are and which AI topics interest you most. What to Know: AI private equity is here For decades, big private equity companies have rinsed and repeated a profitable strategy: buy a struggling firm, trim its expenses by firing staff or modernizing business practices, and then sell it, before moving on to the next. It was only a matter of time before somebody modified that strategy for the age of AI. OpenAI announced on Monday it had taken a stake in Thrive Holdings, a company set up earlier this year by Joshua Kushner’s Thrive Capital, which has a plan to buy and turn around struggling companies by enhancing their productivity with AI. Traditional private equity has been good for C-suite executives, but less so for the average worker. Many employees at PE-acquired companies complain of their new bosses making decisions that might boost balance sheet numbers in the short term but make no long-term sense, or paying themselves large bonuses while firing staff down to a skeleton crew. Now it seems OpenAI will be directly involved in a company following in the footsteps of this grand tradition. AI in Action DeepSeek released a new version of its V3 model on Monday, which, according to some benchmarks, is on par with GPT-5 and Gemini 3.0 Pro. Expect as a result more hand-wringing in Washington and Silicon Valley about the degree to which Chinese AI is catching up to its American rival. What We’re Reading “OpenAI blames boy’s suicide on ‘misuse’ of its technology,” by Robert Booth in The Guardian Great form from OpenAI’s lawyers here: According to filings at the superior court of the state of California on Tuesday, OpenAI said that “to the extent that any ‘cause’ can be attributed to this tragic event” Raine’s “injuries and harm were caused or contributed to, directly and proximately, in whole or in part, by [his] misuse, unauthorised use, unintended use, unforeseeable use, and/or improper use of ChatGPT”.