Palo Alto Networks to buy Applied Compute in $4.5B all-stock deal
Palo Alto Networks agreed to acquire open-weight AI platform Applied Compute in an all-stock deal worth $4.5 billion, its eighth acquisition this year.

Palo Alto Networks has agreed to buy Applied Compute, an open-weight AI model company, in an all-stock deal worth $4.5 billion. It's the eighth acquisition the cybersecurity vendor has announced this year, and it's a notable one. Applied Compute isn't a security product. It's a platform for building and running AI models, and Palo Alto wants to put security around that process.


Why a security vendor wants a model platform
Here's the thing: the enterprise model question has changed faster than most security roadmaps. Companies are no longer just consuming hosted models through an API. Many want to fine-tune open-weight models on their own data, serve them internally and control where that data flows. That creates a new attack surface, and it's one that sits outside the traditional perimeter.
Palo Alto Networks is betting that whoever runs the post-training and inference layer will be well placed to secure it. According to the companies, Applied Compute offers enterprises a platform for post-training and inference on open models. That's the piece Palo Alto says it wants, and it's the reason the deal makes sense on paper.

What the deal includes
The terms are straightforward. Palo Alto would acquire Applied Compute's platform and its research team in an all-stock transaction valued at $4.5 billion. The companies expect the deal to close in Palo Alto's fiscal third quarter of 2027, which covers February to April.
Palo Alto says the technology will let customers train and deploy models securely, whether they run them in a public cloud or on-premises. For IT teams, that second option matters. Plenty of regulated and data-heavy organisations want model training and inference kept close to their own infrastructure, and any security layer has to work in those environments too.

Applied Compute CEO Yash Patil will become Palo Alto Networks' co-chief AI officer when the deal closes. Palo Alto noted that the title isn't currently held by anyone at the firm, so it's effectively a new role.
What Applied Compute brings
Applied Compute was founded in 2025 and has grown quickly. The startup reported that its team size tripled from January, and that 85% of its staff are researchers and engineers. That's a heavily technical headcount, and it's the main asset Palo Alto is buying alongside the platform.

The company's valuation has also moved fast. Applied Compute was valued at $3.25 billion after raising $350 million in early November. Nvidia, Blackstone and Lux Capital are among its backers. Customers named in coverage of the deal include Microsoft, Shopify and DoorDash.
Patil framed the pitch in terms of accessibility. He said the company built its product "to make it easy for any company to train, serve, and improve proprietary models," and that joining Palo Alto would "put security at the center of all our products." That's the vendor's own framing, and it's the pitch Palo Alto will need to deliver on once the teams are combined.

Palo Alto's acquisition pace
Eight deals in one year is a fast clip, even for a company that has been buying steadily. Recent additions include Clutch Security, Koi Security, Portkey, Console and Embrace. Portkey's AI gateway and Clutch's work on AI agents show a pattern: Palo Alto is assembling tools that sit around AI usage, not just around the network.
That pattern is worth tracking. Integration is where many of these deals get messy. Bringing a research-heavy startup into a large security company tends to create friction over product roadmaps, release cadence and who owns customer support. How Palo Alto handles that will say more about the strategy than the price tag does.

The security argument, and what Palo Alto is claiming
CEO Nikesh Arora put the strategic logic plainly. "Every enterprise is going to run a mix of open and closed models, and every one of those models needs to be secured," he said. He also pointed to Palo Alto's own evaluation work. Palo Alto's Unit 42 team carried out a Frontier AI Assessment of Applied Compute, which Arora said "confirmed their strong security foundations." He added that the company had seen the platform first-hand as a prospective customer.
Those are the company's claims, and they're worth reading as such. A vendor assessing an acquisition target is not an independent audit. What matters for practitioners is whether the combined product delivers controls for model supply chains, inference endpoints and data handling that fit into existing security operations.
What to watch before the deal closes
The deal won't close until early 2027, so there's time to plan. If you run AI workloads on open-weight models, this is a signal to review how models are sourced, where they're trained and served, and which logging and access controls cover them. Teams in this position typically find that model governance sits in a gap between data science, security and infrastructure, and the acquisition may push vendors to fill that gap.
Watch three things in the coming months: how Palo Alto describes the product roadmap for on-premises deployments, whether existing Applied Compute customers keep their current terms, and how the co-chief AI officer role is defined once Patil starts. Those details will show whether this is a platform play or a feature bolted onto a security suite. And yet, the direction is clear: security vendors want to own the layer where enterprise models are built and run, and they're paying for it.