Palo Alto Networks will buy Applied Compute, a startup that helps companies customize and run AI models, for $4.5 billion in stock. This Palo Alto Networks acquisition was announced Wednesday and the price may change through closing adjustments.
The security vendor announced it alongside quarterly earnings that beat last year’s by a wide margin.
The product Palo Alto is buying
Applied Compute sells software that takes open-weight models, which anyone can download, and tunes them on a customer’s private data so they can run in production. Palo Alto Networks wants to sell that to its own security customers, deployed in public clouds or in sovereign environments. The startup’s current customers will keep being served.
The deal pushes Palo Alto Networks past its core job of defending AI systems and into helping clients build them. CEO Nikesh Arora predicts companies will mix open-weight and proprietary models. He said Palo Alto first considered Applied Compute as a customer, and that its Unit 42 threat team reviewed the platform’s security before the deal.
People and timing
Yash Patil, who co-founded Applied Compute and runs it as CEO, becomes co-chief AI officer at Palo Alto Networks after closing. The startup’s researchers move over too, and they will train in-house models for the company’s products.
Closing is targeted for the company’s third fiscal quarter of 2027, if regulators approve and the usual conditions are met.
This is the second AI purchase in a few months. In September it bought Console, an AI-native product for building agentic workflows and automating operational work through plain-language instructions.
Earnings announced the same day
For the fiscal first quarter, which ended October 31, Palo Alto Networks reported:
- Revenue of $3.44 billion, up 39% from a year ago.
- Next-generation security annual recurring revenue of $9.73 billion, up 66%.
- GAAP net income of $482 million ($0.57 per diluted share), against $334 million ($0.47) a year earlier.
It also lifted guidance. Fiscal 2027 revenue is now forecast at $14.52 billion to $14.57 billion, a 26% to 27% increase.
Why a security company wants a model platform
Companies that tailor open-weight models to their own data worry about where that data goes and who can reach the models. A vendor that supplies both the platform and the protection can sell the two together, and that is the logic behind this Palo Alto Networks acquisition. The company hasn’t published pricing or product details for a combined offer.
The risk is integration. Applied Compute is a young company, and merging its team and customers into a large public company takes time. Buyers who chose the startup for its independence may want clear answers on how their data and models are handled afterward.
For existing customers nothing changes right away, since the startup keeps operating as it does today.
What security buyers get
Many large buyers are weighing whether to run models inside their own cloud accounts instead of sending data to a model vendor. Deployment in public clouds and sovereign environments speaks to customers with data-residency rules, such as governments and regulated industries.
It also opens a second revenue line for Palo Alto Networks. Today it sells the tools that protect AI deployments. With Applied Compute it can sell the deployment tooling as well, and Arora’s view that both open-weight and proprietary models will be used suggests the company wants to serve either path.
The research team’s role is a quieter signal. Training in-house models for use across products suggests Palo Alto Networks wants its own AI capability rather than only reselling someone else’s.
The deal value is described as subject to adjustments, so the final figure could differ from $4.5 billion. Because payment is entirely in stock, the cost to Palo Alto Networks also moves with its share price until closing, a point investors usually track on deals of this type.
Taken with the Console purchase in September, the pattern is clear: two AI-focused acquisitions within about two months.
What to watch through fiscal 2027
Look for regulatory approval, the planned close in the third fiscal quarter of 2027, and the first products that bundle the platform with Palo Alto’s security tools. The company also needs to explain what in-house models its new researchers will train. Details come from Palo Alto Networks’ Wednesday announcement.
Source: SecurityWeek