OpenAI's Trillion-Dollar Gamble: Balancing Massive Spending with Ambitious Revenue Plans

OpenAI has five years to turn $13 billion into $1 trillion | TechCrunch

OpenAI is printing money right now. The company is pulling in roughly $13 billion in annual revenue, with 70% coming from everyday people paying $20 a month to chat with an AI, according to the Financial Times. That's pretty wild when you consider ChatGPT has 800 million regular users, but only 5% are actually paying subscribers. Raking in billions though it may be, OpenAI has also committed to spending over $1 trillion over the next decade (yes, trillion). The company has recently locked in deals for more than 26 gigawatts of computing capacity from Oracle, Nvidia, AMD, and Broadcom -- infrastructure that'll cost vastly more than what's coming in. To bridge this gap, OpenAI is getting creative, reports the FT. A five-year-plan include exploring government contracts, shopping tools, video services, consumer hardware, and even becoming a computing supplier itself through its Stargate data center project. A growing number of businesses need to math to work out. Some of America's most valuable companies are now leaning on OpenAI to fulfill major contracts, notes the FT; if OpenAI falters (no pressure!), it could potentially destabilize a U.S. market that has banked heavily on the AI boom.

OpenAI makes five-year business plan to meet $1tn spending pledges

OpenAI is working on new revenue lines, debt partnerships and further fundraising as part of a five-year plan to make good on the more than $1tn in spending it has pledged to create world-leading artificial intelligence. OpenAI is planning on deals to serve governments and businesses with more bespoke products, creating more income from new shopping tools, and new sales from its video creation service Sora and AI agents, said multiple people familiar with the start-up's efforts. These people said it is exploring "creative" plans to raise new debt that can help it build out its AI infrastructure, while considering becoming a supplier of computing resources via its data centre initiative Stargate. It is also weighing ways to cash in on its intellectual property by developing new AI infrastructure, making forays into online advertising and plans to launch consumer hardware products, including a new AI-powered personal assistant device, with former Apple star designer Jony Ive. These ambitious plans will need to become reality if OpenAI is to meet its liabilities, as the group has made funding commitments that dwarf its income. In the past month, chief executive Sam Altman has committed to take more than 26 gigawatts of capacity from Oracle, Nvidia, AMD and Broadcom, at a rough cost of well over $1tn over the next decade, according to FT calculations. The ability to meet these costs is increasingly a concern for the wider economy. Some of the most valuable companies in the US are now reliant on OpenAI to fulfil major contracts and underpin demand, stoking fears of an AI-fuelled financial bubble. One senior OpenAI executive said "[investors] expect you to have a five-year model", but added "right now I'd say there's lots of fuzz on the horizon, and as it gets closer and it's going to start to take real shape". OpenAI books about $13bn in annual recurring revenue, 70 per cent of which comes from consumers using ChatGPT, which costs $20 for a standard subscription, according to people familiar with the company's finances. ChatGPT has more than 800mn regular users, but just 5 per cent of those are paying subscribers, a number OpenAI intends to double, the senior executive said. The company has also rolled out cheaper access to users in India, with plans to do the same in the Philippines, Brazil and elsewhere, they said. It also takes a cut of sales from items purchased through ChatGPT's new checkout feature and is exploring introducing advertising to its AI products. Altman last week said he liked Instagram's approach to personalised advertising: "Maybe there's something to do there, but, we approach ads with great caution." Recent partnerships with AMD and Nvidia include plans to share "technical expertise" in order to improve AI hardware, including chip and data centre design. One executive at the company compared those plans to Jeff Bezos launching cloud computing platform AWS using technical expertise gleaned from running his ecommerce business Amazon. OpenAI's operating loss in the first half of the year was about $8bn, even as revenue more than doubled on the year before, said a person with knowledge of the matter. The company's partners such as Oracle have taken on the upfront spending on infrastructure, with OpenAI hoping it can grow to meet its obligations to those partners as operational expenditure in future. The approach has been to "leverage other people's balance sheets" to give OpenAI "time to build the business", said the senior executive. Greg Brockman, the company's president, last week said recent spending commitments would pay for themselves: "If we had 10 [times] more compute [computing power], I don't know if we'd have 10 [times] more revenue, but I don't think we would be that far." If OpenAI continues its stratospheric growth, executives are also confident they can keep raising money from investors. Alternatively, the start-up could prioritise breaking even, though Altman last week said becoming profitable was "not in my top-10 concerns". OpenAI is also anticipating computing costs will fall sharply as a result of competition among suppliers and technical advancements. The company's deals with AMD and Nvidia are staggered so OpenAI will pay as new capacity is developed. But 20GW of capacity would require power roughly equivalent to that provided by 20 nuclear reactors, and analysts have questioned whether it is realistic for that demand to be met by a single company. Two-thirds of the cost of developing new computing power goes towards semiconductors. OpenAI is aiming to stimulate the nascent chip financing market by offering enormous demand, and by forging novel contracts, such as its Nvidia and AMD deals. Those deals have been criticised for their circularity -- the ChatGPT maker is expected to spend much of Nvidia's investment on the chipmaker's processors, for instance. But the transactions will help non-investment grade OpenAI raise the debt it requires to realise its infrastructure ambitions, said the senior executive at the start-up. The signal to the market is "we're good for the debt", they added. "We're working with everyone to come up with creative financing strategies." "I don't view them as drunken sailors going to bars and laying down IOUs everywhere," said a person who has advised the company on its dealmaking. "It might look and feel that way, but this is actually a strategy backed up by technology, products, business plans and visibility into what is happening."

A guide to the $1 trillion-worth of AI deals between OpenAI, Nvidia and others

Many predict that the artificial intelligence boom will dramatically change how people live and work, and the scale and pace of recent AI deals seems to reflect this. At the center are a handful of companies that are increasingly turning to each other to finance and build out the necessary infrastructure. ChatGPT-maker OpenAI alone has racked up around $1 trillion in deals this year, according to a report from the Financial Times. In September, OpenAI confirmed that it would pay Oracle $300 billion for computer infrastructure over the course of five years. This deal is part of a $500 billion data center buildout project called Stargate, to which Japan's SoftBank Group is also contributing. OpenAI has also inked a $22 billion deal with CoreWeave for use of its data centers, which are packed with Nvidia graphics processing units. Most recently, OpenAI announced a partnership with Broadcom to develop and deploy racks of chips designed by the AI startup. That deal amount is undisclosed. OpenAI has been able to go on this shopping spree because of a $100 billion investment from Nvidia, though a large portion of that money likely be used for leasing Nvidia's GPUs. Microsoft has also invested about $14 billion in OpenAI since 2019. Nvidia has also spun a similar web of intertwining AI deals. Nvidia in September agreed to pay up to $6.3 billion for any of CoreWeave's unsold cloud-computing capacity through 2032. CoreWeave gets most of its GPUs that it then rents out to customers from Nvidia, which is also an investor in the AI cloud infrastructure company. Meanwhile, Oracle has purchased about $40 billion worth of Nvidia chips to build a data center for OpenAI, which is part of the Stargate project. Softbank has a $3 billion stake in Nvidia. Some experts are worried that these inflated AI company valuations "are at a bubble." A recent report by Bain & Company found that AI companies will need $2 trillion of annual revenue to fund the infrastructure needed to meet projected demand for AI by 2030. That's an $800 billion shortfall. But AI leaders are pushing back against concerns, saying that this is simply what it takes to make AI a reality. "The largest tech companies in the world are purchasing this infrastructure because they have demand," CoreWeave CEO Mike Intrator said on Mad Money on Oct. 8. "There's nothing circular about that. It's a fundamental infrastructure buildout that's taking place, and when you have such a massive-scale investment in infrastructure, it is not unusual to see partnerships as people try to serve infrastructure to the consumer. It happens in other markets, it's happening in this one." Watch the video for a visual representation of this entangled web of AI deals.

Tracking OpenAI's deals with Nvidia, Walmart, AMD, and more

OpenAI has entered its contract era. The company behind ChatGPT has stitched together chip supply at an unprecedented scale, widened its cloud footprint, and taught a chatbot how to close a sale. The through-line isn't mystique but mechanics: lock down what's scarce, rent what's flexible, and turn an audience into a checkout line. The scope finally matches the talk. A letter of intent with Nvidia puts at least 10 gigawatts of systems on OpenAI's roadmap, with up to $100 billion of Nvidia investment tied to the rollout. AMD arrives as a true second source -- with a warrant that could hand OpenAI up to 10% of AMD if milestones are hit -- while Broadcom signs on to build OpenAI's first in-house processor. Add Google Cloud as a supplier, and the single-vendor optics fade. The other lane for OpenAI is revenue. ChatGPT's "Instant Checkout" started with Etsy, moved to Shopify next, and is now part of a partnership with Walmart. Customers can now complete a purchase directly inside the ChatGPT interface. Stripe provides the payment infrastructure, and OpenAI earns a transaction fee on each sale. That moves ChatGPT from demo to storefront and, if the conversion math holds, it turns intent into income without sending users back to a browser. Money is following the build. As of June, OpenAI's annualized revenue run rate recently hit about $10 billion, nearly double December's pace. Last year's losses were heavy; the bet is that unit costs fall as capacity lands and that commerce and enterprise channels widen the margin. OpenAI is now the world's most valuable startup after a secondary stock sale pegged its valuation at $500 billion. OpenAI spent 2024 laying the groundwork for this year's sprint. Apple announced opt-in ChatGPT access across iOS, iPadOS, and macOS as part of Apple Intelligence; Microsoft relinquished its OpenAI board-observer seat under regulatory scrutiny; Oracle, Microsoft, and OpenAI said Azure AI capacity would be extended over Oracle Cloud Infrastructure; PwC became OpenAI's first ChatGPT Enterprise reseller and bought around 100,000 seats; and OpenAI signed licensing deals with the Financial Times, News Corp, and Reddit to allow attributed answers and data access. The question right now isn't whether OpenAI can sign deals. It can. And it will continue to do so. The question is whether these contracts buy real leverage -- on delivery windows, on price curves, on distribution -- fast enough to make the economics sing.

OpenAI earns $13 billion a year while planning a trillion-dollar AI future

To support its long-term growth, OpenAI has secured deals for more than 26 gigawatts of computing capacity from suppliers including Oracle, Nvidia, AMD, and Broadcom. OpenAI is generating approximately $13 billion in annual revenue, largely from user subscriptions, while concurrently committing to over $1 trillion in spending over the next decade to expand its computing infrastructure and capabilities. According to a report from the Financial Times, 70 percent of the company's revenue is derived from individuals paying $20 per month for chat access. Although ChatGPT has a base of 800 million regular users, only 5 percent of them are paying subscribers who contribute to this revenue. To support its long-term growth, OpenAI has secured deals for more than 26 gigawatts of computing capacity from suppliers including Oracle, Nvidia, AMD, and Broadcom. The cost of this infrastructure investment is expected to vastly exceed the revenue currently being generated by the company. To bridge this financial gap, OpenAI has formulated a five-year plan that involves diversifying its operations. The strategy includes pursuing government contracts, developing shopping tools, creating video services, and producing consumer hardware. The company also intends to become a computing supplier itself through its planned Stargate data center project. Several of America's most valuable companies now rely on OpenAI for the fulfillment of major contracts. The Financial Times noted that a significant failure at the AI firm could potentially destabilize the broader U.S. market due to this growing dependency.