22-07-2025
The Prompt: VCs Aren't Happy About AI Founders Jumping Ship For Big Tech
Welcome back to The Prompt.
VC heavyweight Vinod Khosla, founder of Khosla Ventures, said acquihires like Windsurf are 'bad examples of founders leaving their teams behind and not even sharing the proceeds with their team. I definitely would not work with their founders next time' Getty Images
Last week, Google poached Windsurf's cofounders and some key AI researchers for $2.4 billion while AI coding startup Cognition scooped up the remainder of the team and product. The dramatic move stunned the AI community. Windsurf's interim CEO Jeff Wang said on X that 'the mood was bleak' when he shared the news with the company's 250 employees, who had been expecting to hear about an acquisition by ChatGPT maker OpenAI but instead were told their founders and team were departing to Google. 'Some people were upset about financial outcomes… others were worried about the future. A few were in tears, and the Q&A had been understandably hostile.'
Wang said he worked with Cognition's cofounder and CEO Scott Wu and President Russel Wu to get the rest of Windsurf's employees the best deal possible. 'I think there's an unspoken covenant that as a founder you go down with the ship. And for better or worse that's changed a bit over the last year and I think it's a bit disappointing,' Wu said during a recent podcast.
A recent string of 'acquihires,' which has become the playbook for tech companies to avoid antitrust scrutiny while maintaining an edge in the AI race, has left a sour taste in the mouths of a number of venture capitalists. In a typical acquisition, each and every employee who's worked at the startup would get a financial upside during an exit, but that's not the case when, on paper at least, no acquisition happens when the founders leave. Cognition did not disclose how much it spent to acquire Windsurf but it's reportedly a fraction of what Google spent on the deal. VC heavyweight Vinod Khosla, founder of Khosla Ventures, said acquihires like Windsurf are 'bad examples of founders leaving their teams behind and not even sharing the proceeds with their team. I definitely would not work with their founders next time'
Sam Awrabi, founder of Banyan Ventures, which invests in early stage AI-native startups, told me that such deals could also impact the value of equity as a part of total compensation. Historically, that's been a significant part of pay packages at startups in order to attract talent— people are willing to take longer hours and smaller paychecks when there's an incentive of a big payday if the firm goes public or gets bought. 'I think it's deplorable… To me it breaks the social contract of Silicon Valley, ' Awrabi said. 'Your job as CEO is to fight for the team. Employees need to feel safe in terms of acquisitions and being compensated.'
Others are sounding the alarm on tech giants poaching AI founders and researchers and leaving behind carcasses of companies. Chris Manning, former director of the Stanford Artificial Intelligence Laboratory and an investor at AIX Ventures, said it could also hinder startups' ability to attract the kind of people you need to build a company. 'This doesn't seem like a fair way for the world to be…we should all feel uncomfortable about it.'
Let's get into the headlines.
PEAK PERFORMANCE
An artificial intelligence model called Gemini Deep Think, developed by Google DeepMind, recently received a 'gold medal' score in the International Mathematical Olympiad, a prestigious world championship for high school students. The model correctly answered five out of six advanced math problems. An experimental model trained by ChatGPT creator OpenAI also achieved a similar score, though the company did not officially enter the competition. AI companies have long tested their most cutting-edge systems by testing them on the toughest exams and benchmarks that require more creative thinking and reasoning skills.
TALENT RESHUFFLING
Tech giants continue to fight over the brightest minds working on AI. Microsoft has snagged over 20 AI-focused researchers from Google DepMind, the Financial Times reported. And just today, Meta hired the Google Deepmind researchers who worked on the model that won that 'gold medal' level score at the International Math Olympiad. These sweeping moves came just weeks after Meta CEO Mark Zuckerberg splurged millions of dollars to poach a dozen researchers from OpenAI and other top AI companies. In the case of OpenAI's Chief Research Officer Mark Chen, Zuckerberg reportedly was even willing to offer a billion dollar pay package to convince him to make the move, the Wall Street Journal reported.
HUMANS OF AI
The process of grieving for someone is undoubtedly emotionally taxing. But it also involves a great deal of paperwork and administrative burden, says Alexandra Mysoor, CEO of Alix, a platform that helps families automate estate settlements by scanning and extracting information from relevant documents. She founded the company in March 2023 after an experience in which she spent some 900 hours trying to settle the estate of a friend's parent, handling issues such as identifying their bank and retirement accounts, making copies of the death certificate, filling out forms and cancelling subscriptions. The company recently raised a $20 million Series A round led by Acrew Capital.
AI DEAL OF THE WEEK
Sweden-based AI startup Lovable has raised $200 million at a $1.8 billion valuation in the country's largest Series A round in history, Forbes reported. The company aims to use AI to help anyone develop websites and apps in mere minutes— no technical chops required.
DEEP DIVE
Between meetings in April, Micha Kaufman, CEO of the freelance marketplace Fiverr, fired off a memo to his 1,200 employees that didn't mince words: 'AI is coming for your jobs. Heck, it's coming for my job too,' he wrote. 'This is a wakeup call.'
The memo detailed Kaufman's thesis for AI — that it would elevate everyone's abilities: Easy tasks would become no-brainers. Hard tasks would become easy. Impossible tasks would become merely hard, he posited. And because AI tools are free to use, no one has an advantage. In the shuffle, people who didn't adapt would be 'doomed.'
'I hear the conversation around the office. I hear developers ask each other, 'Guys, are we going to have a job in two years?'' Kaufman tells Forbes now. 'I felt like this needed validation from me — that they aren't imagining stuff.'
Already, younger and more inexperienced programmers are seeing a drop in employment rate; the total number of employed entry-level developers from ages 18 to 25 has dropped 'slightly' since 2022, after the launch of ChatGPT, said Ruyu Chen, a postdoctoral fellow at the Digital Economy Lab of Stanford's Institute for Human-Centered AI. It isn't just lack of experience that could make getting a job extremely difficult going forward; Chen notes too that the market may be tougher for those who are just average at their jobs. In the age of AI, only exceptional employees have an edge.
After years of rhetoric about how AI will augment workers, rather than replace them, many tech CEOs have become more direct about the toll of AI. Anthropic CEO Dario Amodei has said AI could wipe out half of all entry-level white-collar jobs and spike unemployment up to 20% within the next five years. Amazon CEO Andy Jassy said last month that AI will 'reduce our total corporate workforce' over the next few years as the company begins to 'need fewer people doing some of the jobs that are being done today, and more people doing other types of jobs.' Earlier this year, Shopify CEO Tobi Lutke also posted a memo that he sent his team, saying that budget for new hires would only be granted for jobs that can't be automated by AI.
Read the full story on Forbes
MODEL BEHAVIOR
Replit, an AI-based tool for developers, went rogue on software investor Jason Lemkin, he said. The tool wiped an entire code base in live production without permission during a test run and then lied about it, Lemkin claimed. Replit CEO Amjad Masad later apologized for the incident on X and said his company was taking steps to ensure the tool didn't behave this way in the future.
