Sam Altman, the CEO of OpenAI, is performing a precarious balancing act. The creator of ChatGPT, one of history's fastest-growing companies, is heading into a decisive and potentially perilous year in 2026. Leaked financial projections paint a stark picture: the artificial intelligence pioneer is expected to burn through a staggering $17 billion in cash during 2026, a sharp increase from an estimated $9 billion in 2025.
The High-Stakes Fundraising Circus
To fuel this astronomical spending, Altman's cap remains perpetually outstretched. Having already raised over $60 billion from investors since ChatGPT's late-2022 explosion, OpenAI is not done. Sources indicate the company will "almost certainly" seek another massive capital infusion in 2026, with reports pointing to a target as high as $100 billion. This could value the company at around $830 billion, up from $500 billion in October 2025.
Potential investors are lining up. Amazon is in separate talks to invest up to $10 billion, a move facilitated by OpenAI untangling its exclusive cloud partnership with Microsoft, Amazon's rival. Even chipmaker Nvidia has expressed willingness to invest up to $100 billion in increments to secure sales of its crucial hardware. Despite Altman occasionally downplaying the idea, rumours of an eventual public offering also persist.
The Core Problem: Revenue Coupled to Costly Computing
This unprecedented fundraising has driven unparalleled growth. Revenue reportedly soared from $1 billion in 2023 to $13 billion in 2025, hitting an annualised rate of $20 billion by year's end—a feat that took Google and Facebook years longer. However, the fundamental challenge remains: OpenAI's biggest cost, computing power, grows in lockstep with its revenue.
The company's computing needs have exploded from 200 megawatts in 2023 to 1.9 gigawatts (GW) in 2025. It has plans to add a further 30GW of capacity in the coming years, a commitment costing an eye-watering $1.4 trillion. The brutal economics were highlighted in a November 2025 leak of Microsoft figures, which suggested OpenAI's costs for running its models (inference costs) actually exceeded its revenue in the first half of the year.
Mounting Competition and Strategic Pivots
Competition is intensifying, squeezing OpenAI's ambitions. According to Stanford benchmarks, the performance gap between top AI models has narrowed significantly. Google's Gemini 3, launched in November 2025, outperformed OpenAI's GPT-5.1 on many metrics, forcing a rapid counter-punch with GPT-5.2. Meanwhile, freely available "open models" are also closing the performance gap.
There are signs of market saturation. A Deutsche Bank study found consumer subscriptions for ChatGPT in major European countries "ground to a halt" in summer 2025. By mid-December, ChatGPT had 910 million monthly active users compared to Gemini's 345 million, but the latter is gaining ground. Recognising the threat, Altman declared a temporary "code red" in December 2025, pausing initiatives like ad integration to refocus on improving ChatGPT's core product.
To diversify, OpenAI is expanding aggressively. It has built a consulting division for enterprise clients, launched tools like AgentKit for work automation, and is experimenting with e-commerce, allowing companies like Walmart to sell products through ChatGPT for a fee. Mirroring Google's playbook, it is also pursuing vertical integration, partnering with Broadcom to design custom chips and hiring Apple's legendary designer Sir Jony Ive to work on a consumer device.
Investor Jitters and a Make-or-Break Moment
Some investors are growing wary of the scale of losses, which one venture capitalist compared to the deficits of national governments. When pressed by loyal investor Brad Gerstner on how OpenAI would fund commitments roughly 100 times its 2025 revenue, Altman testily retorted, "If you want to sell your shares, I'll find you a buyer." Critics draw parallels to the collapse of WeWork, calling it "the WeWork story on steroids."
Altman argues that economies of scale will eventually improve as training costs become a smaller fraction of total revenue. Yet, with training costs still mounting and rivals at its heels, 2026 is shaping up to be a defining year. The company must prove it can monetise its technology beyond consumer subscriptions—through enterprise sales, advertising, or new hardware—while managing a cash burn rate that threatens to undermine its spectacular growth. The future of the AI lab that started it all hangs in the balance.
