The deal. Palo Alto Networks (PANW) has signed a definitive agreement to buy Applied Compute. Applied Compute sells a platform for post-training (tuning an existing AI model on a company's own data) and inference (actually running the model) for open-weight models, meaning models whose weights anyone can download and host. The price is $4.5 billion, paid all in stock and subject to adjustments. It still needs regulatory clearance and is expected to close in Palo Alto's fiscal third quarter of 2027. Co-founder and CEO Yash Patil becomes Palo Alto's Co-Chief AI Officer at close.
Why Palo Alto wants it. The company says most enterprises will end up running a mix of open and closed models, and that no large provider is delivering open models securely today. The plan is to let customers train, adapt and deploy open-weight models with security built in, in public cloud or in sovereign (in-country, self-controlled) environments. Nikesh Arora said Palo Alto first saw the platform as a prospective customer and that a Unit 42 Frontier AI Assessment checked its security. The research team will also help train Palo Alto's own large and small models.
The quarter. In fiscal Q1 2027 (ended Oct 31), total revenue grew 39% to $3.44 billion, Next-Generation Security ARR grew 66% to $9.73 billion, and remaining performance obligations grew 35% to $21.0 billion. Non-GAAP EPS was $1.05 versus $0.93 a year ago; GAAP EPS was $0.57. CFO Dipak Golechha said the company is "materially raising" full-year guidance; Investing.com reports a new fiscal 2027 range of $14.52 to $14.57 billion in revenue and $4.26 to $4.30 in non-GAAP EPS, and shares up about 7.5% after hours.
What it means for buyers. Palo Alto is no longer just selling tools that guard AI; it wants to host and tune the models too. That puts it closer to cloud and AI platform vendors than to classic security peers. If you run open models in house, expect Palo Alto sellers to pitch model hosting and model security as one package. Watch the earnings call for how the stock deal affects dilution and the AI roadmap.
