OpenAI’s chief financial officer has told colleagues she does not believe the company will be ready for an initial public offering in 2026, directly contradicting CEO Sam Altman’s push for a Q4 listing, according to a report from The Information published Sunday.
Sarah Friar’s concerns center on two numbers: $200 billion in projected cash burn before OpenAI reaches positive cash flow, and $600 billion in cloud server commitments over the next five years. She has questioned whether the company needs to pour that much capital into compute infrastructure, and whether slowing revenue growth can support the obligations, Moneycontrol reported.
The Vendor-Investor Overlap
Friar flagged a structural risk in OpenAI’s recently closed $122 billion funding round: a significant portion of the committed capital comes from Amazon and Nvidia, both of which are also major suppliers of cloud and chip infrastructure to OpenAI. That dual role — investor and vendor — creates what she views as a potentially conflicted capital structure, per The Information’s reporting.
OpenAI is currently generating roughly $2 billion per month in revenue. But with a $200 billion cash burn projection before profitability, the math requires sustained and accelerating revenue growth at a scale no AI company has demonstrated.
Three Executive Signals in Three Weeks
Friar’s concerns land in a compressed window of leadership turbulence at OpenAI. On April 3, COO Brad Lightcap shifted out of the operational role to a special projects position. The same week, applications CEO Fidji Simo took medical leave and Kate Rouch assumed interim responsibilities.
The Information also reported that Friar’s own reporting line has shifted — she now reports to Simo rather than directly to Altman, an unusual structure for a public-company CFO, according to Storyboard18. She has at times been excluded from key financial discussions, including a recent meeting with a major investor on server procurement.
Both executives have publicly said they are aligned on OpenAI’s broader compute strategy. Reuters could not independently verify The Information’s report. OpenAI did not respond to a request for comment.
The Compute Allocation Question
For teams running agent workloads on OpenAI’s API, the $600 billion compute buildout is not abstract. How OpenAI allocates that capacity — between consumer ChatGPT, enterprise contracts, government/defense partnerships, and API access for third-party agent platforms — determines pricing, rate limits, and availability for every developer building on the platform.
If Friar’s caution prevails and spending slows, the compute buildout contracts. If Altman’s timeline holds and OpenAI races toward an IPO under public-market pressure, the incentive shifts toward maximizing revenue per GPU-hour — which historically means prioritizing high-margin enterprise deals over API access. Either scenario changes the economics for agent builders who depend on OpenAI as their inference provider.
The core question has shifted from whether OpenAI can build enough compute to whether the people responsible for the money agree on what the compute is for.