Fiscal first quarter 2027 earnings call, November 18, 2026
A $4.5 billion bet on training and supervising enterprise AI models
Palo Alto Networks announced its intent to acquire Applied Compute for $4.5 billion, paid entirely in stock. It expects to close in fiscal Q3 2027. The target includes roughly $500 million in cash and prepaid compute commitments. Applied Compute specializes in post-training open-weight models and reached more than $100 million in annualized revenue within 18 months of its founding. Palo Alto expects the business to add more than $1 billion of next-generation security ARR by fiscal 2029. CFO Dipak Golechha called it "our next multibillion-dollar ARR business." The company expects it to be top-line accretive from day 1.
CEO Nikesh Arora laid out the logic as a progression. Prisma AIRS first provided discovery of AI assets, then a single enforcement point for AI traffic. After the Hugging Face incident, the company concluded that deterministic controls "cannot anticipate every scenario that an agent is going to conduct." Palo Alto then decided it needed an LLM that can interpret and stop misbehaving AI. Arora said such a supervisory model must be trained alongside the application it watches: "I cannot go in and intercept and understand somebody else's application because I don't know how it's supposed to behave because I didn't train it." In his words, "we have to get in the training business to be able to build a good supervision business." He said the company evaluated 9 providers before choosing Applied Compute, and he expects 100 frontier AI researchers in-house by year-end. He also said that "if it works, I think this will dwarf the ability that we have in cybersecurity, this could be bigger than what we do today."
Arora said the team will be deployed on two fronts. The first is embedding AI across Palo Alto's own products, with more than 35 specialized use cases in development. The second is training and serving secure, sovereign open-weight models for enterprises. He cited open-weight models growing from under 5% of token traffic to more than 60% over the past year, and described a serving opportunity "in the hundreds of billions of dollars."
Financial cost: margin dilution and a new capex cycle
The deal comes with a financial price, which management stated plainly. Assuming a fiscal Q3 close, Palo Alto expects the transaction to lower operating margin by up to 100 basis points in fiscal 2027 and up to 200 basis points in fiscal 2028, before returning to its current 29% to 30% range in fiscal 2029. The company plans to build more than 100 megawatts of diversified inference capacity over 3 to 5 years. Cumulative capex through fiscal 2030 is expected to be less than $3 billion, which Golechha put at about half of this year's expected free cash flow. Most of the spend would go to GPUs and servers. Management expects free cash flow margins of 38% or more excluding data center spend in fiscal 2028 and beyond. Guidance excludes Applied Compute.
Analyst John DiFucci noted the deal is very different from Palo Alto's existing business. Arora said the company can serve initial demand through pay-as-you-go and 3-year compute contracts already in place, and that 100 to 150 megawatts can come from a neocloud or a cloud provider. He said Palo Alto would only need to become a compute specialist at gigawatt scale. Fatima Boolani asked whether the move erodes Palo Alto's neutrality as a "Switzerland" security vendor to AI infrastructure players. Arora argued the end-to-end capability does not exist in the industry yet, and said "before you start building a Switzerland product, let's make sure your Apple product works well." He added that he expects others to follow.
Results: broad acceleration and a record NGS ARR beat
Next-generation security ARR reached $9.73 billion, up 66%, and net new NGS ARR more than doubled. Management described it as one of the largest NGS ARR beats in company history. Revenue rose 39% to $3.44 billion. RPO grew 35% to $21.0 billion and current RPO grew 36% to $9.4 billion, with contract duration flat. Non-GAAP EPS was $1.05, up 13% and $0.08 above guidance, and GAAP EPS was $0.57. Adjusted free cash flow was $1.9 billion, a 56% margin. There were no share repurchases, with $1 billion of authorization running through December 2027. By region, the Americas grew 42%, EMEA 40% and JAPAC 26%.
Network and AI security revenue grew 20% to $2.3 billion. SASE ARR rose 30% with more than $100 million of net new ARR. Trailing 12-month firewall bookings were up 28%, and software firewall ARR grew 34%, accelerating for a third straight quarter. Cortex revenue grew 53%, or 40% pro forma. XSIAM ARR rose over 65% to about $800 million. Observability ARR grew about 4x, including roughly 9 figures of net new ARR from a large frontier lab customer migrating from an incumbent. Identity security revenue grew 23% pro forma, with more than 100 net new logos from the existing installed base. Golechha said Idera operating margin is on track to converge with the core business in about half the targeted 18 to 24 months.
Prisma AIRS reached nearly $170 million in ARR across more than 1,000 customers within 5 quarters of general availability, making it the fastest-growing product in company history. Net new platformizations hit about 170, more than the prior two first quarters combined. Platformized customers represent two-thirds of NGS ARR with net retention above 120%. Large deals included $85 million with a global financial institution, which was the largest deal in CyberArk's history. They also included $46 million with a global consulting firm, including the largest Prisma AIRS purchase to date, and $20 million with a global IT services firm....