LE PERSPECTIVE / THE STORY

Palo Alto Networks has agreed to buy Applied Compute for $4.5 billion in stock, taking the cybersecurity company further into the business of building and running enterprise AI, not just protecting it.

Applied Compute specializes in post-training and inference for open-weight models. Its technology helps companies adapt existing AI models to their own needs and run them in production. Palo Alto Networks wants to pair that capability with security designed into the platform, whether customers use public clouds or sovereign environments.

Security Moves Into the Model Layer
Palo Alto Networks wants to secure how enterprise AI models are adapted and run, not just the apps built on top of them. That makes Applied Compute a bet on owning a critical part of enterprise AI infrastructure.

The bigger shift: Palo Alto Networks is betting that security should sit inside the systems enterprises use to customize and deploy AI models, rather than being added after deployment. That could give it a role much earlier in customers’ AI decisions.

Why open-weight AI matters

Palo Alto Networks expects businesses to use a mix of proprietary frontier models and open-weight models. The latter can offer more control over customization, costs and where workloads run, especially for organizations with strict data-handling requirements.

The challenge is that choosing a model is only one part of the job. Enterprises still need to adapt it, serve it reliably, monitor its behavior and address security risks. The acquisition is meant to bring those steps closer together under one provider.

What Applied Compute brings

Applied Compute brings a platform for model post-training and inference, along with a research team that Palo Alto Networks says will work across its wider product portfolio. The buyer also plans to use the team's expertise to train its own large and small AI models.

Yash Patil, Applied Compute's co-founder and CEO, is set to become Palo Alto Networks' Co-Chief AI Officer when the deal closes. The rest of the Applied Compute team is also expected to join the company.

An interesting detail: Palo Alto Networks says it evaluated Applied Compute as a prospective customer, and its Unit 42 team assessed the startup's security foundations. That suggests the interest goes beyond simply adding another AI product to its catalog.

The deal and the next test

The agreement values Applied Compute at $4.5 billion, subject to adjustments, with payment entirely in Palo Alto Networks shares. It still needs regulatory clearance and other customary conditions. The company expects a close in its fiscal third quarter of 2027.

Applied Compute is expected to keep serving AI-native businesses and existing enterprise customers. Palo Alto Networks plans to introduce the platform to its global customer base, which it says exceeds 70,000 organizations.

The acquisition would put a cybersecurity vendor closer to the model-development and deployment workflow itself. The question after closing will be whether Palo Alto Networks can turn that combination into a secure, useful AI platform enterprises actually want to operate.

Get stories like this in your inbox every week. Subscribe to The Week in Perspective.