OpenAI’s Chief Financial Officer Sarah Friar used a November 6 interview to stress that the company will continue expanding its AI infrastructure, even after a wave of local referendums signalled public resistance to new data centres ([1]) ([2]). Voters in 40 of 41 ballot measures across 10 US states backed restrictions on data-centre development this week, reflecting rising community pushback over the environmental and economic impact of the AI boom ([3]). Despite the midterm election backlash, which contributed to a steep sell-off in technology stocks, Friar said OpenAI “certainly will continue to invest in artificial intelligence infrastructure” ([4]). The CFO emphasised that building data capacity quickly is essential to meet growing customer demand for OpenAI’s generative AI services and to support the company’s goal of an eventual IPO ([5]).
OpenAI’s stance highlights a stark divide in the current AI investment landscape. On one hand, it illustrates that leading AI-focused firms remain confident enough in future returns to push ahead with large-scale spending, even when facing regulatory headwinds and cost concerns. On the other hand, outside the tech sector there is still little sign of any AI-driven boost to financial results: as Apollo Global Management’s chief economist recently noted, beyond Big Tech “the technology has delivered zero financial payoff to corporate America ([6])”. This dynamic puts pressure on CFOs at traditional companies to carefully weigh short-term cost pressures against the risk of falling behind on longer-term AI-driven growth opportunities.
One of big tech’s most prominent finance executives is jumping into the AI infrastructure race. David Wehner, Meta’s former CFO and, until recently, its chief strategy officer, announced on 7 November that he will depart Meta to become the chief financial officer of Fireworks, a Silicon Valley startup building AI data-centre technology. The Information and CNBC, which reported the move, noted that Fireworks was valued at $17.5 billion as of a July funding round ([1]). Wehner’s defection underscores the high expectations for companies providing the “picks and shovels” of the AI boom, as major investors pour capital into the computing infrastructure needed to run advanced AI models.
The recruitment of a seasoned CFO like Wehner to a young AI company signals that even cautious finance leaders see substantial long-term value in the AI infrastructure segment. In contrast to mainstream corporations still struggling to quantify returns from their AI projects, only 37% of firms globally report any positive impact on earnings from AI so far ([2]), and a mere 6% meet the threshold for significant bottom-line impact ([3]), companies that enable AI development are attracting outsize valuations and talent. In effect, while many end-user organisations have yet to reap clear financial gains from deploying AI, the expectation of future payoff is driving aggressive investment in the tools and platforms behind AI, where immediate demand and revenue opportunities are more evident.
As enthusiasm for AI gives way to operational realities, finance leaders are increasingly reining in spending on projects that lack clear evidence of value. Deloitte’s third-quarter CFO survey found that many large firms lack the ability to quickly curtail AI usage even when costs exceed approved limits ([1]), and KPMG’s global poll highlighted that only 26% of organisations have full, real-time visibility into what they are spending to run AI models ([2]). This limited cost governance, combined with ballooning cloud and model fees, is forcing difficult choices. An IBM Institute for Business Value study reports that every executive surveyed has already cancelled or delayed at least one generative AI initiative because its operating costs outweighed the benefits ([3]).
In this climate, boards and CFOs are making a decisive pivot from “blank cheque” AI experimentation towards disciplined cost management and proof of return. Finance leaders are now more often the gatekeepers of AI investments, requiring robust business cases and clearly defined performance metrics before signing off on new projects. The emerging lesson is that AI initiatives must show tangible gains, such as productivity improvements or cost savings with accountable owners, to avoid being scaled back. Until a broader swath of companies can point to measurable profit or efficiency gains from AI, spending will likely be focused on high-priority use cases with demonstrable ROI potential.