OpenAI says Stargate, its massive AI infrastructure of data centers, chips, cloud capacity and power, has already passed its initial security goal. 10GW of US AI capacity by 2029. The milestone comes as OpenAI is mulling a possible IPO that Reuters reported could target the maker of ChatGPT. up to $1tn.
For investors, the timing raises a sharp question about the business model. OpenAI is raising the actual cost of running and developing ChatGPT before the public markets fully realize whether the revenue will foot the bill.
OpenAI said it added more than 3GW of capacity in the past 90 days alone, a pace that shows how quickly the company believes demand for AI is growing. The announcement followed reports that OpenAI missed revenue and user growth targets, and concerns within the company about whether future data center costs could be funded if revenue did not grow fast enough.
Stargate is the infrastructure engine behind the next phase of OpenAI. In plain English, it means more data centers, more chips, more cloud capacity and more power so that ChatGPT and future AI tools can train bigger models, answer more users, provide companies and developer support. OpenAI needs momentum now to keep products fast, reliable and competitive. Proof of income comes later. That’s not good for any IPO story because public investors will end up asking how much of each dollar of ChatGPT’s revenue is left after paying for the infrastructure behind it.
A typical software company can usually add users at the top end once the platform is built. Generative AI has a steep cost curve. All rapid, model development, business tool and engineering work uses expensive computing power. More usage can increase revenue, but it can also increase delivery costs.
OpenAI’s previous IPO debate had already focused on whether ChatGPT’s revenue would exceed the costs of running the technology. Stargate is increasing the number because the infrastructure goal has been achieved years ahead of schedule. Energy can create growth, but only if enough users and customers pay at levels that justify construction.
The reported targets make it difficult to sell. OpenAI still has to keep building so that it doesn’t fall behind Anthropic, Google, Meta and other AI competitors. Waiting too long can hurt performance and limit business sales. Creating immediate risks is putting money before profit.
Microsoft’s relationship adds another layer to the IPO case. OpenAI says Microsoft is still its primary cloud partner, while OpenAI can now offer products to any cloud provider. Microsoft’s license for OpenAI models and products continues until 2032, but is now non-exclusive. That gives OpenAI more room to secure capacity than a single cloud provider. It also gives future investors more to evaluate: who provides the infrastructure, who gets paid first, and how much revenue is left after the data center and cloud partners take their share. Stargate’s pure financial review is for use. Securing 10GW of AI capacity sounds strong, but unused or underutilized data center capacity could weigh on margins. Energy sold to high value businesses, developer demand and government demand can make spending easier to protect. Business customers can decide if the numbers work. Consumer use gives OpenAI scale and product power, but businesses often bring large contracts, tight budgets and tight maintenance. If Stargate helps OpenAI sell reliable AI power to companies and governments, the data center bill becomes part of a bigger revenue engine.
Consumer demand is less certain. ChatGPT remains one of the most well-known products in the technology, however the reported lack of user growth suggests that the app may not convert all the waves of social attention into the predictable paid income. A future IPO will need to demonstrate that free use, paid subscriptions and business contracts strike the right balance. The broader market is already facing the same question. Reuters reported that shares in AI-related companies fell after a Wall Street Journal report about OpenAI’s missed targets, with Oracle and CoreWeave among names hit by concerns over OpenAI’s growth prospects. OpenAI’s financing plans are now reaching developers, cloud providers, data center operators and energy providers.
Stargate has also been linked to the creation of a massive AI infrastructure involving OpenAI, SoftBank, Oracle and other partners, with a potential investment of up to $500bn. That scale explains why investors are watching OpenAI’s revenue path closely. If demand for AI continues to rise and customers pay for it, the buildout supports a strong growth story. If revenue disappoints, the sector is left with a large cost base and weak return visibility.
A multi-billion dollar estimate would make the margin of error even narrower. Public investors may accept large spending if it creates long-term profitability, but they will expect evidence that spending translates into quality revenue, customer retention and a path to revenue generation. OpenAI still has strong arguments – ChatGPT is a product for global consumers, the demand for AI is still growing, and additional infrastructure can improve the speed, reliability and performance of the models. That power can support high-value products and large business deals.
The pressure is on for OpenAI to scale the machine before the market fully realizes the profit model. That may be necessary for AI, where volume shapes the product itself, but it also makes the case for an IPO more demanding.
OpenAI has shown that it can secure the infrastructure for the next wave of AI demand. The next proof point is financial: whether ChatGPT’s revenue and business AI can justify the data center bill before the company hits the public markets.
More from Finance Monthly – Anthropic Amazon Compute Deal: What the $100 Billion Commitment Means for AI Infrastructure Power
