OpenAI Enters the Hardware Business, The Challenges and Opportunities for Jony Ive, Takeaways from Google I/O 2025
Summary
OpenAI’s nearly $6.5 billion all-stock purchase of Jony Ive’s io is its clearest bid to own the consumer AI interface rather than remain an app on someone else’s platform. Ben says an acquisition makes more sense than a loose partnership and fits OpenAI’s broader effort to become “the interface for AI.” His strategic framing: with Google certain to compete through models and Android, OpenAI sees an opening to become “the Apple of this space.”
The deal forces OpenAI toward a consequential choice between Apple-like integration and Google-like service ubiquity. Owning hardware could capture the full experience and economics if AI retains meaningful marginal costs, but it will also turn potential distribution partners into competitors and bias OpenAI’s services toward its devices. Ben’s warning from Google’s Android history: “The tail started wagging the dog.”
The $6.5 billion headline buys a concentrated Apple hardware organization, but not Apple’s supply-chain leverage. The operation has recruited former Apple leaders including Evans Hankey and Tong Tong, along with specialists spanning industrial design, manufacturing design, and operations; Ben says they have “plucked the best people.” Yet those veterans are accustomed to Apple dictating supplier terms, while OpenAI must build a smaller organization and perhaps a family of products amid supply-chain decoupling: “Hardware is really hard.”
The product opportunity is real, but neither host believes a new interface automatically displaces phones or chat. Altman called today’s interaction model the “terminal phase” preceding AI’s graphical interface; Ben counters that asynchronous text already mirrors how humans prefer to communicate and keeps users connected to the physical world. Glasses offer an outward camera, with audio potentially handled in the ears—though Ben notes that wearing devices in the ears for hours is problematic—and Andrew doubts any product could persuade non-wearers to adopt them daily.
Ive’s record supplies credibility and a warning that world-class design still needs a ruthless editor. His teams produced the colorful iMac, iPod, MacBooks, iPhone, and iPad, but Ben faults late-era Apple for elevating form over function through products such as the butterfly-keyboard MacBook Pro and $17,000 gold Apple Watch. Steve Jobs was “the world’s greatest editor in chief”; the nearly ten-minute, self-congratulatory launch film suggested that io still needs somebody willing to cut.
Google I/O showed industry-leading AI capability and a persistent inability to turn that capability into standalone consumer products outside Search. Ben praised Google’s models, image generation, and Veo 3, yet struggled to identify how ordinary users would discover the demos; even Google lacks a Gemini app for Mac, while its Flow interface felt rough. His concise diagnosis: Google remains “a one-product company, and that product is search.”
Google’s defensible plan is to use AI Mode as a proving ground, then “graduate” winning features into Search and Android at roughly six-billion-user scale. Real-time data connections, generated visualizations, AI Overviews, Lens, defaults, and existing habits give Google distribution that OpenAI must persuade users to seek out. That scale is simultaneously the moat and constraint: advanced features cannot reach main Search until they are reliable and cheap enough, while the $250 plan primarily attracts highly switchable enthusiasts.
Veo 3’s coherent eight-second clips with audio are already astonishing, and current cost and model capability are the immediate constraints on longer mass-market video. Ben described generation costs as “astronomical,” with each clip consuming substantial credits; as a Gemini Ultra customer, he now has about $550 per month in free APIs. As costs fall, his optimistic response to synthetic-media chaos is that people may finally assume “everything online is fake,” trust physical reality, and “go and touch grass”—a future Andrew hopes for but does not confidently expect.
Deep dive
1. The io acquisition turns OpenAI’s interface ambition into a company
Andrew opened with Bloomberg’s report that OpenAI would acquire io, the secretive startup co-founded by Jony Ive, for nearly $6.5 billion in stock—its largest acquisition in OpenAI’s history. The deal creates a dedicated AI-device unit and brings in former Apple designers; Ive described it as the culmination of 30 years and predicted “products and products and products.”
Ben’s immediate resolution was organizational: partnering indefinitely to build tightly integrated hardware would be awkward, whereas acquisition makes io unambiguously part of OpenAI. It also complements the previously discussed Windsurf deal because both serve the same ambition—OpenAI wants to control “the interface for AI.”
Ben added a pointed corporate-structure observation: OpenAI is transitioning from nonprofit toward for-profit status while the nonprofit board retains control. In his framing, issuing stock to businesses such as Windsurf and io is also “a good way to diminish” how much of the eventual company belongs to the nonprofit.
Hardware matters because consumer capabilities need a tangible route to adoption: “You go to the store, you buy this thing, this thing lets you do things.” ChatGPT’s emergence through a website and app is itself extraordinary—Andrew marveled that, in 2025, so many people voluntarily type chatgpt.com—but new AI demos increasingly need a product that makes them coherent and easy.
2. OpenAI is choosing whether to be AI’s Apple or everybody’s service
Ben treats Google’s presence in AI as inevitable. Its business-model conflict is obvious, but so are “unmatched” infrastructure and an elite research team; he recalled OpenAI’s founding emails in which Elon Musk and Sam Altman argued that Google would capture the field unless somebody acted.
Ben imagined Altman’s systems-level choice this way: Android will be deeply integrated with Google’s models, leaving OpenAI as merely another app—the same position Meta resents on iOS. The alternative is to assume Google owns the broad platform and pursue “the opportunity to be the Apple of this space.”
Full integration may also make economic sense if serving AI never becomes marginally negligible in the way consumer web services are treated. OpenAI could own hardware, models, and experience as an “integrated all-up player,” capturing more value instead of absorbing inference costs while another platform controls the customer.
Andrew’s pushback was that OpenAI appears to want Google and Apple inside one company. Listener Robert preferred advertising and commoditized complements—“become the next Google, not the next Apple.” Ben called that the correct objection: devices make potential partners defensive and cause service development to favor owned hardware, as Google once became so focused on Android versus Apple that “the tail started wagging the dog.”
3. Apple talent travels, but Apple’s supplier power does not
The personnel make this more than a famous designer’s retirement project. Ben highlighted Evans Hankey, Tong Tong, and recruits covering industrial design, the intermediate work of making designs manufacturable, and operations. The team knows Apple’s internal talent map and has systematically “plucked the best people.”
That expertise matters because modern phones are extraordinarily dense systems, not sculptures. A credible device must look good while integrating extensive capability, battery life, reliability, annual iteration, quality assurance, and mass production; this is why Ben rejected easy comparisons with failed projects assembled from former Apple employees. “This isn’t a fly-by-night hardware operation picking up the scraps.”
Starting now gives OpenAI one advantage: it can design its supply chain for the current uncertainty around China rather than unwind an entrenched system. The disadvantage is cultural experience—these employees are used to operating inside the Apple machine, where suppliers accommodate extraordinary volumes and Apple can finance a competitor when it dislikes one vendor’s price.
Andrew stressed that io remains far smaller than Apple’s organization even while discussing a “family” of products. Ben thinks the team may underrate how different its supplier leverage will be, although Apple veterans may be better equipped to navigate those problems than complete newcomers. The balanced call: the pedigree is real, but “hardware is really hard.”
4. Ive’s résumé is unmatched, and his late Apple years remain a warning
Ben’s CliffsNotes version of Ive’s career began before Steve Jobs returned, when the disillusioned designer was close to leaving. Jobs elevated him into a right-hand role, and Ive’s teams produced the colorful teardrop iMac that helped revive Apple, then the iPod, MacBooks, iPhone, and iPad—making him, in Ben’s view, history’s most impactful product designer.
Ben nevertheless remains critical of Ive’s later tenure. Apple’s 2010s Mac design often elevated form over function, culminating in excessively thin machines and the butterfly-keyboard fiasco, although Ben cannot know how responsible Ive was while “halfway out the door.” The $17,000 gold Apple Watch felt overly precious, even though its fundamental design endured and the broader Watch line became successful.
Ive’s essential partnership was with Jobs, whom Ben called “the world’s greatest editor in chief”—the person who knew what to push, pull, improve, or discard. Ive once told Apple interns that withholding criticism is selfish rather than caring because “true care is being honest”; like Gregg Popovich yelling at Tim Duncan, visible coaching of the star gives everyone else permission to demand better work.
5. AI may need a new form factor, but chat has not been displaced
Altman promised consumer-hardware quality that “has never happened before,” prompting Ben’s eyes to roll: Apple deserves respect for delivering remarkable quality at global scale every year. OpenAI can instead compete through superior AI, freedom from cannibalizing an incumbent product, and experiences an established phone maker might hesitate to ship.
The positive interpretation of a new form factor is that OpenAI will not waste time reinventing the phone; beginning as a differentiated phone accessory would be acceptable. The negative interpretation follows Ive’s unease about what smartphones did to society: excessive control designed to prevent bad uses might also eliminate the product’s upside until it is “just not used at all.”
Altman’s “terminal phase of AI interactions” meant text-command terminal, not terminal illness. Ben explained that terminals provide expansive control while graphical interfaces constrain users to whatever controls designers anticipated; generative UI might resolve that tradeoff by creating interfaces dynamically, which is why Altman says AI’s equivalent to the graphical user interface has not yet been found.
Ben still thinks chat is underrated because asynchronous text matches ordinary human communication: ask, leave, and return when the answer is ready. He finds glasses difficult to beat as a form factor because they provide an outward camera, while audio can be delivered through the ears—though he notes that wearing devices in the ears for hours is problematic. Andrew’s pushback was personal but consequential—any device would need to be spectacular before a non-glasses wearer accepts it every day.
6. The launch film exposed the danger of design without an editor
Listener Alex captured the announcement’s backlash: Altman and Ive appeared to be “adopting me against my will from an orphanage,” while the self-flattering, seemingly staged film approached the discomfort of Apple’s hydraulic-press advertisement. Details such as clinking espresso cups and San Francisco streets reportedly closed for filming made the performance of authenticity feel even more engineered.
Ben’s verdict on the roughly nine-minute-and-50-second film was that it should have been three minutes. Andrew found the announcement page equally uncomfortable, while Ben compared its typeface to a wedding invitation. The irony was hard to miss: a partnership centered on elite taste launched with material that appeared to lack precisely the forceful editor Ive once valued.
Ben connected that preciousness to product policy. If Ive remains tortured by the iPhone’s harms, the team may constrain useful behavior until the device lacks appeal; “good and bad are a package,” and individuals ultimately decide how tools are used. He distinguished efforts to ensure AI does not “kill everyone” from a paternalistic “people can’t handle the truth” strain that he finds oppressive.
7. Google won the capability demo and still left a product gap
Ben’s reaction to Google I/O was deliberately split. The keynote offered a “cornucopia of outcomes” downstream from extraordinary teams and infrastructure: he judged Google’s current model the best, called its image and video generation “crazy,” and highlighted the newly convincing ability to render text inside generated media.
Yet the only presentation he found easy to anchor on was Search, because people understand a product they can immediately use. Google can still ship compelling Search features; outside that franchise, Ben saw “this vast gulf” between technical capability and productization. His distilled assessment: “Google is a one-product company, and that product is search.”
The NotebookLM example captured both sides. Bill Bishop used it to summarize a Sharp China episode in fewer than 2,000 words and, Andrew said, “it hit all the right beats.” The unresolved commercial question is how many ordinary consumers will deliberately find NotebookLM, learn the workflow, and repeat it rather than default to something already embedded in their daily behavior.
Ben contrasted OpenAI’s underrated, functional apps with Google’s missing Gemini app for Mac and a rough Flow experience. Apple looked worse still: Whisper made transcription broadly available long ago, yet Apple’s own transcription remains weak. After another account of Apple’s AI failures, Ben encouraged OpenAI to attack.
8. Search and Android give Google a credible AI distribution funnel
Google’s most coherent strategy was what Ben called the AI or Google funnel. Experimental capabilities enter AI Mode, where the interface can resemble ChatGPT and Google can refine them; the best features then “graduate to regular Search” once quality, reliability, and economics permit deployment at mainstream scale.
Google’s data connections make that funnel unusually powerful. A user can ask a sports-related question, receive the answer, and see a visualization generated on demand. Ben called AI Overviews the world’s most-used generative-AI product and framed Google’s race as making Search good enough that people reserve ChatGPT for special work but continue Googling everything else automatically.
Habit is the consumer moat. Google pays for default placement because it can ride Apple’s device dominance without requiring initiative; ChatGPT’s success despite that friction is therefore “a miracle.” Android offers Google the corresponding path for new AI behavior—put Lens and visual search directly on phones, then let users adopt them without hunting for separate products.
Scale is simultaneously Google’s advantage and constraint. Andrew suggested features must become cheap enough for two billion daily users; Ben corrected him to “more like six billion.” He compared all foldable phones sold globally with roughly one day of Apple phone volume. The $250 Google tier can expose frontier capabilities, but its subscribers are enthusiasts likely to switch whenever Anthropic, Gemini, or another model becomes fashionable.
9. Veo 3 makes synthetic video likely to expand before society is ready
Andrew described Twitter filling with nine-second Veo 3 clips; Ben emphasized the almost unbelievable achievement of coherent eight-second video with audio. A year or two earlier, virtually nobody expected this capability, yet reactions already shifted from astonishment to asking why the clips cannot be longer—a pattern he thinks makes confident long-range AI forecasting unreliable.
The immediate constraints are cost and current model capability. Ben called video-generation expense “astronomical”: generating a clip consumes a lot of a Gemini Ultra customer’s credits, and he now has about $550 per month in free APIs. The model is operating near the edge of present capability because it must generate and assemble the entire audiovisual sequence coherently.
Ben nevertheless expects duration to increase and costs to fall through the normal course of technology. Andrew fears an environment where anyone can instantly produce convincing three-minute fabrications; Ben agreed that cheap, broadly available generation deserves concern, even while insisting that today’s eight seconds are already an extraordinary technical result.
Ben’s Formula Fakers example showed the social transition underway: a plausible fake post about Jack Doohan, who had recently lost his seat to Franco Colapinto, made fun of Doohan’s crash and generated a real controversy. His hopeful endpoint is that synthetic media accelerates until everyone assumes “everything online is fake,” trusts physical experience, and goes outside to “touch grass.” Andrew left room for that outcome while calling it an optimistic spin on a potentially dystopian future.