Welcome back to In a loopTIME's new twice-weekly newsletter on artificial intelligence. If you're reading this in a browser, why not? subscribe to have your next email delivered straight to your inbox?
Who to know: Ali Ghodsi, CEO of Databricks
The three most valuable private companies in the US have great reputations: OpenAI, SpaceX and Anthropic. But the fourth, Databricks, remains a little under the radar. The company, which is currently raising $134 billion in funding, according to reports this week, is the quiet workhorse of the artificial intelligence revolution. Databricks offers a platform where companies can combine unwieldy data from different sources, easily manage it, and train AI models with it. Its CEO, Ali Ghodsi, is a charismatic computer science professor who still teaches occasionally at the University of California, Berkeley, even though he runs a multibillion-dollar company. I talked to him last week.
AGI – According to Ghodsey, artificial general intelligence has already been created. To justify this provocative claim, he recalls his experience in a computer science lab two decades ago, when he discussed with colleagues in vaunted terms what AGI might look like: one that would be able to talk, reason, and identify patterns in massive amounts of data. “Now that we have it, it’s kind of like never having met your heroes,” Godsey laughs. “It’s just like, okay, this isn’t that impressive.”
Okay, but “Godsey has good reasons for this argument.” While OpenAI is one of its largest clients, Databricks' bread and butter comes from a long tail of more conventional companies that aren't doing AI but just want to use existing AI to work with their data. According to Ghodsey, Databricks is capitalizing on this opportunity. “If all the progress in artificial intelligence has been frozen today, I think we have what it takes to continue what we're doing,” says Ghodsey. “We're not trying to build a supergod… We're much more focused on how can we make AI useful in organizations today?”
About AI agents –“Agents” was the buzzword of 2025, but it never delivered. If anything, AI has improved radically this year, including new capabilities to perform web searches or modify and run code. But large AI models still face reliability issues that limit the amount of time they can operate autonomously, and we don't yet have full AI peers. Ghodsey says Databricks is working to solve this problem by helping companies train small, laser-focused agents using their own data. These models end up being cheaper to run and more robust for specific tasks than cutting-edge AI models like ChatGPT or Claude. “We can give you good performance on a specific task because we cheat and make boring AI, focusing only on the specific tasks you need help with,” Ghodsey says.
If you have a minute, please use our quick survey to help us better understand who you are and what AI topics interest you most.
What you need to know: Private investment in artificial intelligence is here
For decades, large private equity firms have washed and repeated a lucrative strategy: buy a struggling firm, cut its costs by laying off employees or revamping business practices, and then sell it before moving on to the next one.
It was only a matter of time before someone changed this strategy in the age of AI. OpenAI announced Monday that it has acquired a stake in Thrive Holdings, a company launched earlier this year by Joshua Kushner's Thrive Capital, which has a plan to buy and turn around struggling companies by boosting their performance using artificial intelligence.
Traditional private equity methods benefit senior executives, but less so for the average worker. Many employees at PE-acquired companies complain that their new bosses are making decisions that may improve the balance sheet in the short term but don't make sense in the long term, or are paying themselves big bonuses while laying off employees down to a bare bones team.
It now appears that OpenAI will be directly involved in a company following in the footsteps of this great tradition.
AI in action
On Monday, DeepSeek released a new version of its V3 model, which is on par with GPT-5 and Gemini 3.0 Pro in some tests. As a result, expect more debate in Washington and Silicon Valley over the extent to which Chinese artificial intelligence is catching up to its American rival.
What we read
“OpenAI blames boy's suicide on 'misuse' of its technology“, Robert Booth in The Guardian
Great form from OpenAI's lawyers here:
According to documents filed Tuesday in California Superior Court, OpenAI said that “to the extent that any 'cause' can be attributed to this tragic event, Rain's injuries and harm were caused or contributed to, directly and proximately, in whole or in part, [his] misuse, unauthorized use, unintended use, unintended use and/or improper use of ChatGPT.”
