Palo Alto Networks CEO Nikesh Arora warns AI firms face a structural 'business model trap' driven by high compute and integration costs

Posted before the market opened. It's not about the NASDAQ Joe. It's about rapid adoption and bringing enterprises along. Lot of software infrastructure needs to be build to make it easier to deploy and sustain AI in enterprises.

So you feel 50-60bn of revenue is going to vanish for the frontier models? As they migrate or you feel it will be distributed and they will get less enterprise market share?

Two down days for the Nasdaq and we start getting posts like this

If open source is good and free - how many will bite the bullet and use it for use cases where they aren't worried about upgradability or total security?

@ppgosavi @nikesharora Building loops with continuous improvement, guardrails and evals is not an easy task for most enterprises (except for very large ones). Hiring the talent to do RSI is also expensive. The AI is already freeing up entry level clerical work - can this talent be utilized as FDEs?

Great thread @nikesharora. Cheap tokens unlock pilots, but deployment needs verifiable outputs. How do you see FDEs evolving to deliver deterministic guardrails + explainability at scale? Will this shift model labs toward more “consulting-like” roles (owning workflows + edge-case training) rather than pure model providers? Curious if verifiable AI becomes the real Phase 2 moat over raw capability.

@nikesharora Wtf? How is using open weights model somehow a compromise for "total security"?

@nikesharora In your opinion, what will be a true cash cow for them in the next 5-10 years? Consumer ads or Enterprise with FDEs? Or both?

More than half of FAANG makes high $ through ads

@TheStalwart ‘The market is behaving normally’ is an editorial position, Joe.

@TheStalwart "Dear Claude, please replace all — with -"

@nikesharora @grok what is the FDE acronym here?

@TheStalwart Can you go back to your old profile picture please

@nikesharora Yup, hold or restrict free AI to consumers, divert all that tokens for enterprises, make AI work for enterprises, earn some money. From that money, give away some to consumers. Not vice-versa currently.

@TheStalwart And to image he was Masa’s heir

@nikesharora If you are right about consumer distribution businesses such as, Google, why haven't their models gotten better than OpenAI and Anthropic models? Initial success of LLMs for coding shows that you don't need FDEs for enterprise deployment if users are sophisticated. 1/2

The pricing point is right, but it solves the wrong half. Cheap tokens unlock experimentation. They don't unlock deployment.

A CIO isn't restricting AI because tokens cost too much. They're restricting it because a confident wrong answer in a real workflow is a liability someone has to sign for. That's a trust problem, not a pricing one.

Which means the paradox might not be one. Refusing to ship a system that can't show its work is the rational call. The Phase 2 unlock is verifiable output: a source on every claim, and "I don't know" as a real answer instead of a fluent guess. Cheap tokens get you the pilot. Only trust gets you to production.

@nikesharora I have a slightly diff POV looking at enterprise apps on a daily basis. Enterprises will go the open source/open model (GLM 5.2) with RSI (recursive self intelligence) as enterprises will take every single % of improvement and lower cost (cant be counting tokens)

@nikesharora 1/2 Similar things have happened before:

High quality FDEs + Domain experts + Agent/workflow/UI builder platform = Enterprise/Model Labs' new best friend India is uniquely positioned to produce a world class company here imo considering FDE talent/cost advantage and tribal business domain expertise via IT services heritage

@TheStalwart LOL