A Security Vendor Buys a Model-Tuning Platform

Palo Alto Networks announced on 18 November 2026 an all-stock deal valued at $4.5B for Applied Compute, a platform for post-training and inference of open-weight AI models. Closing is expected in the company's fiscal third quarter of 2027 and depends on regulatory clearance. Applied Compute co-founder and CEO Yash Patil becomes Co-Chief AI Officer at Palo Alto. The buyer says every enterprise will run a mix of open and closed models and that each one needs securing, and it wants customers to train, adapt and deploy open-weight models in public or sovereign clouds. Applied Compute will keep serving its current customers.

The Quarter That Reported the Same Day

Palo Alto reported its fiscal first quarter of 2027 on the same day, and CNBC says shares rose more than 7% afterwards and are up about 85% this year. Because the price is paid in stock, the real cost to Palo Alto moves with that share price. CEO Nikesh Arora told analysts the new business could outgrow cybersecurity if it works, which is a bet and not a result.

MeasureFiscal Q1 2027Change
Revenue$3.44BUp 39% from $2.47B
Next-generation security ARR$9.73BUp 66%
Net income$482MUp from $334M
Full-year revenue outlook$14.52B to $14.57BRaised above $14.16B expected

What to Ask If You Use Either Company

If you are an Applied Compute customer, ask now for written terms on neutrality toward other security vendors, pricing after closing, where your training data and model weights are stored and how you can leave, because nothing changes before the deal closes but the roadmap can change afterwards. If you buy from Palo Alto, expect a bundled offer for securing AI models and ask what it costs on top of your current contract. The stated aim covers sovereign clouds, so European buyers should ask in which region tuning jobs run before they send regulated data.