Differences have emerged between OpenAI's Chief Financial Officer Sarah Friar and CEO Sam Altman regarding the company's timeline for going public. While Altman has committed $600 billion in spending over the next five years and privately indicated a desire for an initial public offering as early as the fourth quarter of this year, the CFO has expressed significant reservations.
Behind the scenes, Friar has voiced concerns that highlight the contradictions and risks inherent in the CEO's highly aggressive expansion plan. According to a person familiar with her views, Friar told colleagues earlier this year that she believes the company will not be ready for an IPO by 2026. She cited the need to complete substantial process and organizational work and pointed to risks associated with massive spending commitments. Friar also expressed uncertainty about whether OpenAI truly needs to invest such vast sums in AI servers in the coming years and whether slowing revenue growth can support these financial pledges.
It remains unclear if OpenAI's recent announcement of $122 billion in total investment commitments has alleviated some of her concerns. This funding, to be delivered in stages primarily from Amazon and NVIDIA—companies that provide cloud servers and chips to OpenAI—is part of a series of circular financial arrangements among major AI firms.
Publicly, the 40-year-old Altman and 53-year-old Friar typically present a united front, as was the case during an investor dinner at Altman's San Francisco home earlier this year. However, since Friar joined OpenAI in June 2024, differences in their priorities have created noticeable tension among close collaborators. Sources indicate that Altman has occasionally excluded Friar from certain financial planning meetings. For instance, in recent months, he discussed server expenditures with a top investor's executives without Friar's participation, an absence described as conspicuously awkward by an attendee. Another person present at a high-level OpenAI meeting noted that Friar was not invited to a session involving major financial decisions, marking a significant deviation from standard practice.
In large corporations, CFOs almost universally report directly to the CEO. However, in August of last year, Friar broke from this norm and began reporting to Fidji Simo, the head of applied AI at OpenAI, rather than directly to Altman. Simo recently informed staff she is taking a short medical leave.
Altman and Friar issued a joint statement through their spokesperson: "We are fully aligned that stable access to computing power is central to OpenAI's strategy and a key differentiating advantage for our expansion. Over the past year, we have both been directly involved in all major compute decisions. The recent $122 billion financing secures our ability to scale compute massively, positioning us to become a core infrastructure layer in AI. This advantage will translate into sustained leadership in research and products, empowering individuals and businesses worldwide to innovate freely."
"A Challenging Role" Altman and Friar have contrasting personalities and backgrounds. Altman has cultivated an image as a world-changing visionary in the mold of Steve Jobs or Elon Musk. Friar, a former Goldman Sachs equity analyst, previously handled finance at Salesforce, assisted with the IPO of Square (now Block), and served nearly six years as CEO of Nextdoor before being replaced amid business struggles. She joined OpenAI to lead fundraising and reassure investors, with Altman describing OpenAI as "the most capital-intensive company in history."
A person who has worked closely with both remarked, "Friar's job is difficult. She serves a founder with immense ambition who wants to push spending limits as far as possible." Such friction is not uncommon in Silicon Valley, where growth-obsessed visionary founders often clash with more cautious financial executives focused on preparing for a public listing.
A notable precedent is Airbnb, where CEO Brian Chesky clashed with then-CFO Laurence Tosi over business strategy and organizational structure, leading to Tosi's departure in early 2018. This created a significant leadership vacuum ahead of Airbnb's IPO and triggered an exodus of executives frustrated by delayed equity monetization. Chesky successfully took the company public two years later and now serves as an informal advisor to Altman.
OpenAI's leadership has experienced prior fractures. In late 2023, two executives briefly ousted Altman before he returned following negotiations. Earlier, several key executives split from Altman to found competitor Anthropic.
Despite the recent massive funding round, Altman and Friar must continue persuading investors to commit hundreds of billions, if not trillions, of dollars. Pressure is mounting to meet revenue and user targets to support the escalating burn rate. Anthropic has emerged as a major threat, surpassing OpenAI in sales of AI models to enterprises and app developers, while Google's Gemini is eroding ChatGPT's dominance in the consumer chatbot market.
Although OpenAI recently raised its five-year revenue forecast by 27%, the company privately warned investors in February that its cash burn by 2030 could be more than double previous projections. Furthermore, people familiar with the matter revealed that OpenAI's gross margin last year fell short of expectations because demand for chatbots and models exceeded forecasts, forcing the company to procure computing power at premium prices—a challenge Anthropic also faces.
Engaging with Goldman Sachs on IPO Plans Despite Friar's reservations about an accelerated IPO timeline, OpenAI has begun laying the groundwork. It has retained law firms Cooley and Wachtell, Lipton, Rosen & Katz and held informal discussions with IPO bankers from Goldman Sachs and Morgan Stanley. Significant shareholders Amazon and NVIDIA could also influence the timing. Altman has privately expressed a desire for OpenAI to go public before Anthropic, which is reportedly planning to discuss its own listing in the fourth quarter.
If pursued, the IPO would rank among the largest in history, presenting a formidable task for Friar's finance team and other departments. OpenAI has already committed billions of dollars years in advance to fund data centers specifically designed for training new AI models.
Publicly, Friar has not suggested she feels pressured by Altman. In an interview last November, she stated, "Sam does give me a lot of pushes and asks... sometimes it does feel like 'there's a lot to do.' But it's more motivating than anxiety-inducing because I came here believing we are doing something unprecedented." At the time, she also noted that an IPO was "not on the roadmap currently," as OpenAI was still adapting its operations to its current scale of expansion.
Many OpenAI employees believe the aggressive server investment plan will help maintain its competitive edge. After being contacted for comment on this article, Friar posted a photo on LinkedIn with her finance team, congratulating them on closing the $122 billion funding round. She wrote, "Thanks to the vision and decisive action of Sam Altman and President Greg Brockman, this forward-looking approach lays the foundation for our long-term success."
Cloud Computing Risks Altman argues that securing more computing power will enable the development of superior technology and ensure sufficient capacity for a growing user base. Executives have long complained that compute constraints have forced delays in launching new products and features.
After Friar became CFO, Altman sought to reduce reliance on leasing chips from cloud providers by announcing a joint venture with Oracle and SoftBank to build $500 billion in data center capacity in the U.S. However, plans for OpenAI to operate its own data centers were shelved after lenders proved unwilling to finance a company with an unproven business model and annual burn rate in the billions for a project of that scale.
Subsequently, under Altman's direction, OpenAI entered into agreements to lease servers from cloud providers like Microsoft, Oracle, and Amazon, committing to over $600 billion in spending over the next five years. Friar explained in an earlier interview, "OpenAI chose to partner with cloud providers to keep the balance sheet lighter."
Even so, OpenAI has made multi-year funding commitments for custom data center projects, a departure from traditional cloud contracts where customers pay on-demand or commit to annual minimums, with providers like Amazon building the infrastructure. Friar noted that the multi-year construction cycle for AI data centers creates inherent timing pressure, necessitating early capacity reservations.
In a recent OpenAI podcast, she said, "There's a clear timing mismatch. I'm making decisions now for compute in 2028, 2029, 2030, not just 2026 or 2027. If we don't place orders and signal data center construction now, there will be no compute available later." According to reports, OpenAI has agreed to share certain economic risks, like cost overruns, with Oracle in its data center deals—terms rarely accepted by cloud customers.
Pointed Criticism from a Rival OpenAI's $600 billion spending commitment is not entirely inflexible. Two people familiar with the compute contracts said the company could slow some data center construction to reduce expenditure or exit certain agreements.
In September last year, Friar recruited Mike Liebenthal, former CFO of Musk's xAI (now merged into SpaceX), to work with her and Brockman on server-related financial deals.
Friar's previously expressed spending concerns align with views from Altman's key rival, Anthropic CEO Dario Amodei. In a February podcast, Amodei stated, "Even if technology advances as quickly as expected, we can't precisely predict how fast it will drive revenue. But data center investments have long lead times. If the timing is off by a few years, the consequences could be catastrophic." He added, "Even a one-year miscalculation... or growth at five times per year instead of ten, could bankrupt a company," in remarks that appeared to allude to OpenAI. "I get the sense some peers haven't seriously modeled their finances and don't understand the risks they're taking."
Comments