Palo Alto Networks to Acquire Applied Compute in $4.5 Billion AI Security Bet
Share
Palo Alto Networks has agreed to acquire Applied Compute in an all-stock deal valued at $4.5 billion, marking a significant strategic shift towards securing and hosting generative artificial intelligence (AI) for the enterprise.
The acquisition, announced alongside the company’s fiscal first-quarter 2027 results, aims to provide organisations with a platform to train, deploy, and secure open-weight AI models. These models can be hosted in the public cloud or within dedicated sovereign environments to meet strict data-residency requirements.
Applied Compute specialises in the post-training and inference of open-weight models. As part of the transition, Applied Compute co-founder and CEO Yash Patil will become Palo Alto Networks’ co-chief AI officer. The startup’s research team will be integrated across the company’s product portfolio to build native AI capabilities.
The Rise of Supervisory AI
Nikesh Arora, Chairman and CEO of Palo Alto Networks, stated that the company intends to help enterprises adopt AI securely regardless of whether they use open-weight models, closed frontier models, or a hybrid approach. Arora noted that open-weight model implementations have surged, now accounting for more than 60% of token traffic according to internal company figures.
A central component of this strategy is the development of “supervisory AI” models. Arora argued that traditional deterministic security controls are insufficient for autonomous agents. To secure these applications, enterprises must train a secondary AI model to supervise the primary system’s behaviour.
“I cannot go in and intercept and understand somebody else’s application,” Arora said during an earnings call, explaining why the company believes it must be involved in the training and serving process to provide effective security.
Investing in GPU Capacity and Sovereignty
To support these new capabilities, Palo Alto Networks plans to build significant compute capacity. CFO Dipak Golechha confirmed that capital expenditure will be directed primarily toward GPUs and servers. The company aims to establish more than 100 megawatts of compute capacity over the next three to five years.
While the company already offers some sovereign security services, such as a European Cortex service governed by Deutsche Telekom, analysts have raised questions about how this will extend to GPU infrastructure. Sanchit Vir Gogia, chief analyst at Greyhound Research, noted that a US-headquartered operator faces complex jurisdictional challenges when offering sovereign hosting.
The deal follows Palo Alto Networks’ previous $25 billion acquisition of CyberArk, which integrated identity security into its platform. By adding model hosting and training to its existing firewall and Security Operations Centre (SOC) offerings, the company is positioning itself at the centre of the entire AI stack.
However, the move toward “vertical integration”—where a security vendor also provides the infrastructure it protects—has drawn scrutiny. Analysts suggest that while one team managing both the model and telemetry can identify threats faster, it also creates a single point of failure if that vendor is breached or faces operational issues.
The transaction is expected to close in Palo Alto’s fiscal third quarter of 2027, which runs from February to April.



