The leaders of OpenAI and Google have been living rent-free in each other’s heads since ChatGPT caught the world by storm. Heading into this week’s I/O, Googlers were on edge about whether Sam Altman would try to upstage their show like last year, when OpenAI held an event the day before to showcase ChatGPT’s advanced voice mode.
This time, OpenAI dropped its bombshell the day after.
OpenAI buying the “io” hardware division of Jony Ive’s design studio, LoveFrom, is a delightfully petty bit of SEO sabotage, though I’m told the name stands for “input output” and was decided a while ago. Even still, the news of Ive and Altman teaming up quickly shifted the conversation away from what was a strong showing from Google at this year’s I/O. The dueling announcements say a lot about what are arguably the world’s two foremost AI companies: Google’s models may be technically superior and more widely deployed, but OpenAI is kicking everyone’s ass at capturing mindshare and buzz.
Speaking of buzz, it’s worth looking past the headlines to what OpenAI actually announced this week: it’s paying $6.5 billion in equity to hire roughly 55 people from LoveFrom, including ex-Apple design leaders Evans Hankey, Tang Tan, and Scott Cannon. They’ll report to Peter Welinder, a veteran OpenAI product leader who reports directly to Altman. The rest of LoveFrom’s designers, including legends like Mike Matas, are staying put with Ive, who is currently designing the first-ever electric Ferrari and advising the man who introduced him to Altman, Airbnb CEO Brian Chesky. OpenAI’s press release says Ive and LoveFrom “will assume deep design and creative responsibilities across OpenAI.”
When LoveFrom’s existing client work is wrapped up, Ive and his design team plan to focus solely on OpenAI while staying independent. OpenAI, meanwhile, already has open “future of computing” roles for others to join the io team it brought over. One job listing for a senior research engineer says the ideal candidate has already “spent time in the weeds teaching models to speak and perceive.” (Total compensation: $460K to $555K plus equity.)
The rough timeline that led up to this moment goes as follows: Altman and Ive met two years ago and decided to officially work on hardware together this time last year. The io division was set up at LoveFrom to work with a small group of OpenAI employees. OpenAI and Laurene Powell Jobs invested in the effort toward the end of 2024, when there were quiet talks of raising hundreds of millions of dollars to make it a fully standalone company. (The OpenAI startup fund, which is bizarrely not owned by OpenAI, also invested around this time.)
Importantly, Ive ended his consulting relationship with Apple in 2022, the year before he met Altman. That deal was highly lucrative for Ive, but kept him from working on products that could compete with Apple’s. Now, Ive and Altman are teaming up to announce what I expect to be a voice-first AI device later next year.
Early prototypes of the device exist (Altman mentioned taking one home in his promo video with Ive). Altman told OpenAI employees this week that it will be able to sit on a desk or be carried around. Supply chain rumors suggest it will be roughly the size of an iPod Shuffle and also be worn like a necklace.
Like just about every other big hardware company, Ive and Altman have also been working on AI earbuds. Altman is set on bundling hardware as an upsell for ChatGPT subscriptions and envisions a suite of AI-first products that help lessen the company’s reliance on Apple and Google for distribution. With his Apple relationship in the rear-view mirror, Ive now seems set on unseating the company he helped build.
Google, meanwhile, was firing on all cylinders this week. AI Mode in Google Search is being rolled out widely. Its product strategy is still disjointed compared to OpenAI’s, but it’s starting to leverage the immense amount of personal data it has on people to differentiate what Gemini can do. If Gemini can hook into Gmail, Workspace, YouTube, etc., in a way that people want to use, it will likely keep many people from shifting to ChatGPT — just like Meta did to Snapchat with Stories in Instagram.
After meeting with Google employees up and down the org chart, I came away from I/O with the feeling that the company doesn’t see a catastrophe on the horizon like a lot of outsiders. There’s a recognition that the ability to buy out distribution for search on Apple devices is probably coming to a close, but Gemini is approaching 500 million monthly users.
ChatGPT is undoubtedly eating into search (it’s impossible to get Google execs to comment on the actual health of query volume), but Google has shown a willingness to modernize search faster than I expected. The situation differs from Apple, which isn’t competitive in the model race and is suffering from the kind of political infighting that Google mostly worked through over the last couple of years.
There’s also no question that Google is well-positioned to continue leading on the frontier of model development. The latest Gemini models are very good, and Google is clearly positioning its AI for a post-phone world with Project Astra. The company also has the compute to roll out tools like the impressive new Veo video model, while OpenAI’s Sora remains heavily gated due to GPU constraints.
It’s still quite possible that ChatGPT’s growth continues unabated while Gemini struggles to become a household name. That would be a generational shift in how people use technology that would hurt Google’s business over the long term. For now, though, it looks like Google might be okay.
- Anthropic couldn’t sit this week out either. The company held an event on Thursday in San Francisco to debut its Claude 4 models, which it claims are the world’s best for coding. With OpenAI, Google, and Meta all battling to win the interface layer of AI, Anthropic is positioning itself as the model arms dealer of choice. It was telling that Windsurf, which is in talks to sell to OpenAI, was seemingly intentionally left out of getting day-one access to the new models. “If models are countries, this is the equivalent of a trade ban,” Nathan Benaich wrote on X. (Also, what does it say about the state of the industry when the supposed safety-first AI lab is releasing models that it knows want to blackmail people?)
- Microsoft Build was overshadowed by protests. There were several interesting announcements at Build this week, including Elon Musk’s Grok model coming to Azure and Microsoft’s bet on how to evolve the plumbing of the web for AI agents. All of that was overshadowed by protestors who kept disrupting the company’s keynotes to protest the business it does with Israel. The situation has gotten so tense that Microsoft tried unsuccessfully to block the ability for employees to send internal emails with the words “Palestine,” “Gaza,” and “Genocide.”
- I tried Google’s smart glasses prototype. I spent about five minutes wearing the reference design prototype of Google’s new smart glasses. They had a small, low-res waveguide in the center of each lens that showed voice interactions with Gemini, a basic version of Google Maps directions, and photos I took. They were… fine? Google knows this tech is super early and that full AR glasses are still years away. In the meantime, it’s smart of them to partner with Warby Parker, Gentle Monster, and Kering to put Android XR in glasses that I expect to start coming out next year. With Apple now planning a similar pair of AI-powered glasses in 2026, Meta’s window of being the only major player in the space is closing.
- YouTube hired Justin Connolly from Disney as its head of media and sports, a move that Disney is suing over.
- Tinder CEO Faye Iosotaluno is stepping down. Her role will now be overseen by parent company Match Group CEO Spencer Rascoff.
- Vladimir Fedorov, a longtime Meta engineering exec, joined Github as CTO.
- Will Robinson, Coinbase’s former VP of engineering, has joined Plaid as CTO.
- Stephen Deadman, Meta’s VP of data protection in Europe, is leaving due to “structural changes.”
If you haven’t already, don’t forget to subscribe to The Verge, which includes unlimited access to Command Line and all of our reporting.
As always, I welcome your feedback, especially if you have thoughts on this issue, an opinion about stackable simulations, or a story idea to share. You can respond here or ping me securely on Signal.
