Palo Alto’s CEO says AI prices must fall 90% before the tech really takes off

Palo Alto Networks CEO Nikesh Arora said the cost of running AI must drop sharply before enterprises deploy at scale, warning that token prices could fall by as much as 90%. He made the comments in response to OpenAI’s claim that its new GPT-5.6 model is 54% […], highlighting continued cost and scaling hurdles. Overall, the message is cautionary for near-term AI monetization timelines.

Analysis

This reads less like a PANW-specific signal and more like a warning that the current AI commercialization curve is still too expensive for most enterprises. The market implication is that revenue ramps for AI application vendors may continue to lag the hype cycle, which compresses multiples for names selling “AI transformation” before the usage economics are proven. In that setup, the near-term beneficiaries are the large platforms that can amortize inference across massive distribution; the losers are standalone software vendors that need immediate token-volume growth to justify premium valuation.

For PANW, the second-order effect is mostly timing, not direction. Cyber budgets are defensive and less sensitive to AI adoption delays, but any AI-security attach opportunity gets pushed out if customers are still testing rather than deploying at scale. That makes PANW relatively resilient versus pure AI software, but it also caps upside from the AI narrative until management can show measurable attach rates or monetization in product gross margin and billings.

The contrarian point is that the market may already agree tokens have to get cheaper; the real question is whether the cost curve falls fast enough to unlock usage before sentiment fades. If model efficiency improves materially over the next 1-3 quarters, today’s skepticism could reverse into a usage boom, which would favor infrastructure and distribution-heavy vendors. Falsifiers are simple: a step-up in enterprise AI bookings, evidence of materially lower inference costs from hyperscalers, or PANW quantifying AI-linked product demand in upcoming earnings calls.

Key Decisions for Investors

• No immediate standalone trade in PANW; treat this as a watch item and wait for the next earnings call to see whether AI-related attach rates or billings are actually moving. Falsifier: management shows measurable AI contribution to ARR or billings acceleration within 1 quarter.
• If expressing the view, prefer a pair: long PANW / short a basket of high-multiple AI application names over the next 1-3 months. The thesis is that cyber demand is budgeted and sticky, while AI app monetization remains price-sensitive; target is modest downside in the short leg with limited PANW beta.
• Use rallies in broad AI software proxies (e.g. IGV or an AI-app basket) to add downside hedges via 1-3 month puts if enterprise AI spend stays anecdotal. Risk/reward is attractive if token-cost headlines keep recurring and guidance revisions follow.
• For PANW holders, consider selling near-dated covered calls into strength rather than adding aggressively here. The comment does not improve the fundamental growth runway enough to justify paying up for an AI narrative premium.