OpenAI valuation talks heat up at $750 billion, as repeated fundraising and infrastructure costs loom in the background

Renewed debate over growth expectations inflating valuation
Questions raised over sustainability of continuous capital raising
Rising pressure from memory and other infrastructure costs

OpenAI has entered discussions to raise new funding at a valuation of $750 billion, drawing intense attention from global capital markets. The sharp jump from valuations seen just last quarter—roughly two-thirds of the current figure—has renewed scrutiny over how OpenAI’s valuation is being set and the broader context of its fundraising.

Repeated capital raises following SoftBank Chairman Masayoshi Son’s investment late last year, combined with mounting costs tied to data centers and server infrastructure, have turned the valuation debate into something far more complex than a headline number.

Limited reflection of standalone performance

According to IT-focused outlet The Information, Sam Altman, CEO of OpenAI, has recently held preliminary talks with multiple investors to raise tens of billions of dollars at a valuation of $750 billion. If completed, the deal would give OpenAI the highest valuation ever among private companies. Just months ago, during an employee share sale in the third quarter, OpenAI’s valuation stood at around $500 billion, meaning its implied value has risen by 50% in a very short period.

Markets are paying close attention to the fact that multiple valuation benchmarks are now coexisting for OpenAI at the same time. The company had been negotiating a $10 billion investment with Amazon based on a $500 billion valuation, but sources say that in more recent fundraising discussions, OpenAI approached investors using the higher $750 billion figure. Presenting sharply different valuations depending on counterpart and timing suggests the number is less a settled price than a negotiating position.

The talks with Amazon offer important context. According to people familiar with the matter, the discussions include OpenAI using Amazon Web Services’ in-house AI chip, Trainium, or expanding its AWS cloud usage. This points to a strategic partnership model built around large-scale infrastructure demand. OpenAI’s valuation, in this sense, is closely tied to assumptions about future cloud consumption, data center construction, and surging compute needs.

As a result, many observers argue that the $750 billion figure reflects not OpenAI’s current operating performance, but a forward-looking bet on the company’s position in the coming AI infrastructure race. Given that partnerships with major cloud providers such as Amazon and Microsoft are a core premise of these talks, OpenAI’s valuation is likely to be shaped more by its ties to big tech capital than by its standalone results.

“Invest first, raise again” model fuels backlash

Skepticism about OpenAI’s valuation remains widespread. The debate dates back to October last year, when Masayoshi Son announced a $500 million investment in OpenAI. At that time, SoftBank’s Vision Fund valued the company at roughly $150 billion, making it the world’s third most valuable startup after ByteDance and SpaceX. From that point on, OpenAI became a focal point of arguments over whether future AI dominance was being priced in far ahead of actual performance.

The problem, critics argue, is that Son’s investment was not an endpoint but the start of another fundraising cycle. Less than six months later, OpenAI was back in the market, with its valuation climbing from $150 billion to $500 billion and now to $750 billion. Each round of capital raising has been accompanied by a sharp step-up in valuation, creating a pattern where external capital flows appear to drive valuation more than underlying growth. This has fueled criticism that OpenAI inflates its valuation with large investments, only to immediately seek more funding.

Concerns are heightened by OpenAI’s cash burn. Industry estimates suggest the company could consume a total of $115 billion by 2029. Adding to that is a $300 billion cloud usage agreement signed with Oracle, which significantly increases fixed cost pressure. Meanwhile, profitability remains distant. Bloomberg projects OpenAI’s revenue at around $60 billion in 2027, raising doubts about whether revenue growth can keep pace with soaring fixed costs.

This structure leaves OpenAI vulnerable to liquidity risk if external funding were to slow. Bloomberg has drawn parallels with WeWork, noting that both companies face high fixed costs without stable, predictable revenue streams. The persistent bubble debate around OpenAI reflects ongoing dependence on fresh capital and chronic cash burn rather than a one-off valuation spike.

Trust-based financing model under strain

So far, OpenAI has relied primarily on equity sales to fund data center expansion, avoiding direct borrowing. Instead, partners and financial institutions take on debt, while OpenAI leases computing capacity. The ongoing “Stargate” project follows this model, with multiple banks extending $38 billion in loans to Oracle and data center builder Vantage, while OpenAI itself stays off the loan agreements. Partners raise funds through special purpose vehicles, keeping debt off OpenAI’s balance sheet.

This approach allows OpenAI to secure massive resources while minimizing reported liabilities. Blue Owl Capital invested $3 billion in OpenAI’s New Mexico data center project, while a banking consortium including Sumitomo Mitsui, BNP Paribas, Goldman Sachs, and Mitsubishi UFJ arranged $18 billion in loans. Including additional financing raised by infrastructure firms such as Crusoe, OpenAI’s partners have shouldered roughly $28 billion in total.

However, the model depends entirely on partner confidence in OpenAI’s long-term success. Loan repayments are backed by long-term lease contracts, and if those contracts falter, lenders can seize land and data center assets. OpenAI avoids direct financial risk, but remains locked into long-term capacity contracts that translate into sustained cost obligations. Altman has openly described leveraging other parties’ balance sheets as a core strategy.

The challenge is that server investment costs are rising rapidly. As data center construction accelerates, prices for high-performance memory and AI accelerators—critical to large-scale computation—are climbing. HSBC estimates that OpenAI could pay a cumulative $702 billion in computing lease fees by 2030, potentially reaching $1.4 trillion by 2033. What began as a strategy to scale compute capacity now tests OpenAI’s ability to manage long-term costs and continually attract new investment.