OpenAI Closes $7B Round at $300B Valuation to Fund Compute Buildout
OpenAI has closed a $7 billion Series F led by SoftBank and several sovereign wealth funds, valuing the company at $300 billion post-money. The capital is designated for compute infrastructure and international data center expansion.
Original sourceOpenAI has finalized a $7 billion funding round at a $300 billion post-money valuation, making it one of the most highly valued private companies in history. SoftBank led the round, joined by sovereign wealth fund participants whose identities have not been fully disclosed. The raise comes roughly a year after OpenAI's previous funding milestone and signals continued institutional appetite for frontier AI exposure despite elevated geopolitical and regulatory uncertainty.
The stated use of proceeds centers on compute infrastructure — GPU clusters, data center build-outs, and international expansion of training and inference capacity. OpenAI's operating costs remain substantial, driven by the scale required to train and serve frontier models. The company has been increasingly vocal about the capital intensity of its roadmap, and this round appears designed to extend its infrastructure lead while it continues its ongoing structural transition from a nonprofit-controlled entity to a public benefit corporation.
At $300 billion, OpenAI's valuation reflects both its current revenue trajectory and significant expectation of future market capture — particularly in enterprise API adoption and the emerging agentic workflow category. The round also reinforces SoftBank's broader bet on AI infrastructure, consistent with its Vision Fund thesis and its separate commitments to U.S.-based AI initiatives announced earlier in 2025. Whether the valuation holds through an anticipated IPO process will depend heavily on OpenAI's ability to convert compute capacity into durable, high-margin revenue.
Panel Takes
The Founder
Business & Market
“At $300B, OpenAI's valuation is pricing in a world where it captures a dominant share of enterprise AI spend AND wins the consumer platform battle AND makes agentic workflows stick — all three have to be true simultaneously. The moat question still isn't answered: if inference commoditizes and API pricing compresses, this capital buys time but not defensibility. I want to see the unit economics on data center utilization before calling this validation rather than a very expensive option on a future that hasn't materialized yet.”
The Skeptic
Reality Check
“SoftBank led this round, which is the same SoftBank that led WeWork's Series H — pattern recognition matters. The $300B number requires OpenAI to become the AWS of AI inference at scale, in a market where Google, Amazon, and Microsoft are all subsidizing competing infrastructure with existing cash flows. The thing that kills this in 24 months isn't a competitor's model — it's that enterprise buyers consolidate AI spend onto platforms they already have procurement relationships with, and OpenAI ends up a premium model vendor in someone else's marketplace.”
The Futurist
Big Picture
“The thesis embedded in this raise is specific and falsifiable: physical compute scarcity will remain a binding constraint on AI capability long enough for whoever owns the data centers to extract monopoly rents. That thesis is riding the trend of model capability scaling with compute, and right now that trend line is still holding — but the dependency is that no architectural discontinuity (sparse compute, efficient inference, synthetic data shortcuts) breaks the scaling assumption before these data centers are paid off. If the international expansion includes jurisdictions with sovereign data requirements, there's a second-order effect worth watching: OpenAI could end up fragmenting its own model into regional variants, which changes the product story entirely.”
The PM
Product Strategy
“The job this capital is hired to do is clear — buy compute headroom so model capability and inference latency don't become the reason enterprise customers churn to a competitor. That's a legitimate product problem, and infrastructure investment is the right solution for it. What I don't see is how international data center expansion maps to a coherent go-to-market motion: localized infrastructure without localized product, compliance, and support is just expensive metal, not a market entry strategy.”