OpenAI is plotting a dramatic arc toward profitability through the end of the decade, but that growing won’t come without some pain. The company reportedly expects to rack up massive annual losses each year, including roughly $74 billion in operating losses in 2028 alone, then pivot to meaningful profits by 2030, according to financial documents obtained by The Wall Street Journal.

The documents, which were shared with investors this summer, reveal an aggressive growth strategy that hinges on massive upfront investment in computing infrastructure, chips and data centers—spending that CEO Sam Altman has described as necessary to meet what he sees as insatiable demand for AI capabilities. The company anticipates burning through roughly $9 billion this year on $13 billion in sales, a cash burn rate of approximately 70% of revenue.

But the financial trajectory only gets steeper before it improves. The documents show OpenAI projects that by 2028, its operating losses will balloon to roughly three-quarters of that year’s revenue, driven primarily by ballooning spending on computing costs. That’s the same year competitor Anthropic expects to break even, according to WSJ.

The numbers underscore the stark divergence between the two most valuable AI startups. While both companies currently burn cash at similar rates relative to revenue, their paths forward split dramatically. Anthropic forecasts dropping its cash burn to roughly one-third of revenue in 2026 and down to 9% by 2027. OpenAI, by contrast, expects its burn rate to remain at 57% in 2026 and 2027.

OpenAI’s plan relies on what amounts to a bet on dominance. The company recently announced it has signed up to $1.4 trillion in commitments over the next eight years for computing deals with cloud and chip giants. It’s spending almost $100 billion on backup data-center capacity alone to cover unforeseen demand from future products and research.

“Demand for AI exceeds available compute supply today,” an OpenAI spokesman told WSJ. “Every dollar we invest in AI infrastructure goes to serving the hundreds of millions of consumers, businesses, and developers who rely on ChatGPT to get more done.”

OpenAI did not immediately respond to Fortune‘s request for comment.

Altman’s strategy for OpenAI requires near-constant fundraising to keep the startup alive, but could backfire if markets cool on AI or its near-term profitability. Investors have already punished tech companies in recent weeks over concerns about AI spending and whether there will be enough revenue to pay for the extensive AI infrastructure buildout.

OpenAI’s financial figures came before it signed its most recent computing deals, meaning the company is likely set to spend even more than the documents suggest. The cash burn is expected to reach $115 billion cumulatively through 2029, according to The Information.

The company’s optimism about an eventual turnaround rests on revenue projections that show explosive growth. OpenAI now expects to reach about $200 billion in annual revenue by 2030, with the company projecting it will turn cash flow positive beginning in 2029 or 2030. Those figures represent substantial increases over earlier projections shared with investors.

OpenAI Chief Financial Officer Sarah Friar said last week the company has healthy margins and could break even if it wanted to. She highlighted the fast growth of OpenAI’s enterprise business, and said the startup was still experimenting with new business models.

Altman has defended the massive infrastructure spending as a strategic necessity.

“We believe the risk to OpenAI of not having enough computing power is more significant and more likely than the risk of having too much,” he posted last week on X.

The contrasting approaches between OpenAI and Anthropic illustrate two distinct philosophies for navigating the AI boom. Anthropic’s costs are growing at a pace more in line with revenue, and the company is focused on increasing sales among corporate customers, which account for about 80% of its revenue. Notably, Anthropic is trying to avoid OpenAI’s costly forays into image and video generation, which require significantly more computing power. Sora 2, OpenAI’s new video-creation app, might be costing the company millions each day.

OpenAI, meanwhile, is diversifying rapidly. The company recently launched Sora 2 and its first web browser, Atlas. It’s also working on a consumer hardware device with Jony Ive’s design company, researching humanoid robots, and looking to add e-commerce and advertising features for ChatGPT. Whether OpenAI’s gamble pays off will depend on whether demand for its products continues to surge at the pace needed to justify the unprecedented spending. The company expects to burn through roughly 14 times as much cash as Anthropic before turning a profit.

For this story, Fortune used generative AI to help with an initial draft. An editor verified the accuracy of the information before publishing.