By: Editorial Team, StoneX Media
Leading frontier AI labs Anthropic carrying a private market valuation of approximately $965 billion and OpenAI close behind at roughly $850 billion, the two most anticipated technology listings in years would arrive at public markets with expectations already built to an extraordinary level. Hyperscalers including Microsoft Azure, Amazon Web Services, Google Cloud Platform, and Oracle Cloud Infrastructure are each committing well over $100 billion in capital expenditure to power an AI buildout whose return profile remains largely unverified. What neither company has been required to disclose is the precise cost of building a frontier model or the per-query economics of serving it at scale. Public listings would change that calculus entirely. The AI IPO wave matters because it gives investors a clearer lens on where the economics of AI actually accrue, and Lee’s view, the biggest winners are still the companies that control the data, infrastructure, workflow, and security layers around the models.
Yi Fu Lee is a Managing Director and Senior Equity Research Analyst at Benchmark, covering AI, cybersecurity, infrastructure, and enterprise software, experienced in a fusion approach of combining fundamental and technical analysis across public and private markets. His direct coverage of top tier enterprise software firms like Microsoft, CrowdStrike, Datadog, Snowflake, and more gives him line-of-sight into how the largest technology companies build and justify capital expenditure commitments, positioning him to assess what prospectus-level disclosures would actually reveal about frontier AI economics.
So far, the cost of training and running frontier AI models has remained opaque, with no public breakdown of development or per-query serving costs required of either Anthropic or OpenAI. A public listing changes that entirely. "At some point you're going to have to tell us a little bit more about the breakdown of the cost. If you're building the most advanced frontier model, you're going to have to break down how much is the training cost to get there, and when the customer asks the questions, how much is the inference cost to generate that," observes Lee. With OpenAI projected to remain cash-flow negative through at least 2028 or 2029, that scrutiny will arrive well before the company has resolved its profitability timeline. For investors, the quality of what gets disclosed in a prospectus may matter more than the headline valuation.
From the perspective of infrastructure providers, the mindset is often "If I don't spend now, I'm not going to have the infrastructure to service these AI companies three to five years later," notes Lee. That logic is shaping capital expenditure decisions across every major cloud platform. Lee's coverage of Microsoft and the broader infrastructure software sector gives him confidence that the largest hyperscalers have genuine customer pipeline visibility to justify their outlay, at least for now. However, he draws a distinction between disciplined spending and competitive pressure to match peers, invoking the Buffett observation that when the tide goes out, firms without sufficient customer bases to support their infrastructure will be exposed.
The AI stack has three value layers consisting of model layer, infrastructure/data layer, and control/security/workflow layer. The earliest signals that AI spending is converting to real demand arrived not from frontier model companies but from the data infrastructure layer. Snowflake raised its annual growth forecast from 27% to 31%, a result Lee notes caught both his desk and the broader sell side off guard, sending the stock up 38% on earnings. Datadog delivered a similar beat on cloud usage, and in his view, "we're seeing early signs of the AI benefit tailwind, and I don't think it has even started yet." Lee identifies cybersecurity as the sector with the most runway remaining, arguing that compliance and governance requirements represent a gating condition on enterprise AI deployment and that firms like CrowdStrike and Palo Alto Networks will benefit increasingly as AI agents multiply the attack surface. AI is revolutionizing the software sector — it is changing the bottleneck from the once mechanical model building or coding process to operationalizing and managing them for the betterment of human.
Access live prices, supply and demand data and actionable market commentary across commodities, equities, currencies and more. Sign up for StoneX Market Intelligence today and receive a 14-day trial.
Sign up for a Market Intelligence trial today See our financial videos hub
--- Written by Gus Farrow, Senior Manager, StoneX TV
--- Expert: Yi Fu Lee, Managing Director, Senior Equity Research Analyst, Benchmark
The subsidiaries of StoneX Group Inc. provide financial products and services, including, but not limited to, physical commodities, securities, clearing, global payments, risk management, asset management, foreign exchange, and exchange-traded and over-the-counter derivatives. These financial products and services are offered in accordance with the applicable laws in the jurisdictions in which they are provided and are subject to specific terms, conditions, and restrictions contained in the terms of business applicable to each such offering. Not all products and services are available in all countries. The products and services offered by the StoneX Group of companies involve risk of loss and may not be suitable for all investors. Full Disclaimer. This content is not intended for residents of any particular country, and the information herein is not advice nor a recommendation to trade nor does it constitute an offer or solicitation to buy or sell any financial product or service, by any person or entity in any jurisdiction or country where such distribution or use would be contrary to local law or regulation. Please refer to the Regulatory Disclosure section for entity-specific disclosures. No part of this material may be copied, photocopied or duplicated in any form by any means or redistributed without the prior written consent of StoneX Group Inc. The information herein is provided for informational purposes only. This information is provided on an ‘as-is’ basis and may contain statements and opinions of the StoneX Group of companies as well as excerpts and/or information from public sources and third parties and no warranty, whether express or implied, is given as to its completeness or accuracy. Each company within the StoneX Group of companies (on its own behalf and on behalf of its directors, employees and agents) disclaims any and all liability as well as any third-party claim that may arise from the accuracy and/or completeness of the information detailed herein, as well as the use of or reliance on this information by the recipient, any member of its group or any third party.
© 2026 StoneX Group Inc. all rights reserved.
Our subscribers have access to comprehensive market analysis from StoneX spanning commodities, equities, currencies and more.
The anticipated public listings of OpenAI and Anthropic would be the first time investors can scrutinize the actual cost of building and running frontier AI models. Yi Fu Lee of Benchmark StoneX explains what disclosure would change and which parts of the AI ecosystem are already signaling real demand.
With OpenAI and Anthropic preparing for public markets, the financial opacity that has defined frontier AI investing is about to end. The same profitability metrics that govern every public software company will now apply to the firms commanding the sector's highest valuations.
When demand for a new stock runs four times the available supply, the price retail investors see is already shaped by a market running in the background. Alex Ridgers of StoneX explains the pre-market mechanics that determine where IPOs actually open and who gets in.
Our market expertise, advanced platforms, global reach, culture of full transparency and commitment to our clients’ success all set us apart in the financial marketplace.
Reach
With access to 40+ derivatives exchanges, 180+ foreign exchange markets, nearly every global securities marketplace and numerous bi-lateral liquidity venues, StoneX’s digital network and deep relationships can take clients anywhere they want to go.
Transparency
As a publicly traded company meeting the highest standards of regulatory compliance in the markets we serve, our financials and record of accomplishment are matters of public record. StoneX’s commitment to “doing the right thing over the easy thing” sets us apart in the industry and helps us build respect, client trust and new partnerships.
Expertise
From our proprietary Market Intelligence platform, to “boots on the ground” expertise from award-winning traders and professionals, we connect our clients directly to actionable insights they can use to make more informed decisions and achieve their goals in the global markets.
