OpenAI is shifting from pure software to a hardware-inflected commercial strategy as it confronts mounting costs, intensifying competition and legal distractions. In a January post, CFO Sarah Friar portrayed a company that has scaled fast — annualized revenue topped $20 billion in 2025, up from about $6 billion in 2024 and roughly $2 billion in 2023 — while weekly and daily user metrics reached new highs. Those gains have been mirrored by a rapid ramp-up in compute: the company’s raw computing capacity grew roughly threefold a year between 2023 and 2025 and by nearly tenfold over that period in total.
The flip side of that growth is an enormous and rising cash burn. External estimates put OpenAI’s cash outflow at about $8 billion in 2025 and project roughly $40 billion by 2028. Microsoft’s 2025 fiscal disclosures signalled the financial strain on partners too: the tech giant said investment losses tied to OpenAI depressed its net profit by about $3.1 billion in a quarter, a disclosure investors translated into an implied quarterly loss for OpenAI in the order of $11.5 billion. OpenAI’s plan to build out a global data-centre network carries jaw-dropping ambitions — some $1.4 trillion of potential capacity investment — but the programme is only around 10% funded so far.
That squeeze helps explain two immediate pivots: the introduction of advertising in free ChatGPT conversations and the move into consumer hardware. OpenAI has begun testing ads that occupy a large portion of the chat interface in its free tier and in a newly launched mid-priced tier dubbed “ChatGPT Go” (about $8), while keeping ads out of higher-priced subscriptions. Evercore and other sell-side analysts envisage advertising as a multi-billion-dollar revenue stream by 2026 and project it could reach tens of billions by 2030, a potential threat to incumbent search-ad revenues.
Alongside ad experiments, OpenAI confirmed it is developing an AI-driven hardware product due in the second half of the year and has hinted at a first September release. The company’s 2023 acquisition of Jony Ive’s hardware startup for $6.5 billion fuelled rumours about consumer devices: leaked speculation ranges from a pen that records handwriting and audio to earbuds, glasses and a screenless smart speaker. The strategic logic is clear. Controlling an endpoint would let OpenAI influence user experience and monetize more directly while hedging dependence on hyperscale cloud providers and partners.
Yet the hardware pivot arrives as OpenAI’s competitive environment tightens. Google’s Gemini models have quickly taken share and, by many benchmark metrics, overtaken OpenAI’s latest systems. Market-share estimates moved from a near-monopoly position for ChatGPT in 2025 to a significantly diminished lead a year later, while Apple announced a multiyear deal to build the next generation of its on-device intelligence on Google’s stack — a loss of distribution for OpenAI at the iPhone level.
Legal turmoil adds another layer of uncertainty. Elon Musk’s high-profile suit accusing OpenAI and Microsoft of breaching the nonprofit origins of OpenAI is scheduled for trial in the spring; Musk seeks damages in the tens of billions of dollars. Even if legal analysts regard a catastrophic judgment as unlikely, the litigation diverts executive attention and complicates fundraising and valuation in an already capital-intensive business.
Taken together, the picture is of a company racing to broaden revenue lines and control distribution at precisely the moment when rivals are improving rapidly and capital demands are peaking. Ads and hardware could materially change OpenAI’s economics if they scale; they could also dilute focus and require fresh capital in a market that now prizes both technical leadership and disciplined unit economics. For investors and partners, the critical questions are whether OpenAI can convert engagement into higher-margin recurring revenue, reduce dependency on external compute, and do so before a funding window narrows.
The coming year will be a test of execution. Delivering a credible hardware product that differentiates on integration, privacy and usability against Apple and Google is difficult and slow; monetizing ads without eroding trust in generative answers is delicate. Meanwhile, backing from Microsoft and other investors will need to bridge a substantial funding gap if the company is to pursue an owned-infrastructure future. How OpenAI reconciles those trade-offs will determine whether its hardware bet is a clever hedge or an expensive distraction.