Palo Alto Bets $4.5 Billion on Securing Open AI Models

Human Verified

Share to: , and

Security operations centre with glowing shield outlines and network screens in a dark hall

Palo Alto Networks has agreed to buy Applied Compute, a startup that helps companies run open-weight AI models securely, in an all-stock deal worth $4.5 billion. The cybersecurity giant announced the agreement on November 18, alongside results that beat expectations.

The Deal

Applied Compute builds a platform for post-training and inference of open-weight models, the stage where a general model is adapted to a company’s own data and then served in production. Its AC2 platform, introduced in August, handles fine-tuning, serving and feedback from use.

The startup was founded in 2025 by former OpenAI researchers Yash Patil, Rhythm Garg and Linden Li, and it raised $350 million at a $3.25 billion valuation this month. Its customers include Microsoft and DoorDash, and it says its revenue run rate has passed $100 million.

The deal is expected to close in Palo Alto Networks’ fiscal third quarter of 2027, subject to regulatory clearance and other customary conditions. Patil will become Palo Alto Networks’ co-chief AI officer after the close, and the startup’s roughly 70 employees will join the company.

Why Palo Alto Is Buying

Palo Alto Networks argues that most enterprises will run a mix of open and closed models, and that every one of those models needs securing. It says no large provider is delivering open-weight models securely today, and it intends to be the one that does.

Chief executive Nikesh Arora said Applied Compute has built the best post-training and inference platform his team has seen, and that its research team is exceptional. Palo Alto Networks had examined the startup as a prospective customer before deciding to buy it outright.

The announcement came with quarterly results that beat expectations: revenue of $3.44 billion, up 39% year over year, and next-generation security annual recurring revenue of $9.73 billion, up 66%. Palo Alto Networks raised its full-year revenue forecast, and its shares closed up almost 10%.

Our Read

This is a security company buying its way into the model-serving stack. Palo Alto Networks is betting that the next enterprise AI purchase is not a chatbot or a copilot, but the plumbing that lets companies run adapted models safely, and it wants to own that plumbing.

The price shows how scarce the talent is. Applied Compute was valued at $3.25 billion in its latest funding round, employs about 70 people and reports a $100 million revenue run rate, yet Palo Alto Networks is paying $4.5 billion.

It is buying a research team and a head start, not a revenue stream. The risk is integration: the team must stay, the platform must fit Palo Alto Networks’ products, and regulators must clear a deal that concentrates AI security talent.

Watch the rivals. Hyperscalers already sell model hosting with their own security, and Palo Alto Networks itself was recently in talks over a smaller agent-security deal. The question is whether customers trust a security vendor, or their cloud provider, with their models.

Why It Matters

If the deal closes, the largest pure-play cybersecurity company becomes a major seller of enterprise AI infrastructure, not just its guard. That changes the competitive map for open-weight adoption in regulated industries, where security sign-off has been the main blocker.

For enterprises, the pitch is simple: adopt open models without building the security layer themselves. For Palo Alto Networks, the test is whether a $4.5 billion bet on 70 people and a young platform turns into a lasting business, not just a headline.

Background

Palo Alto Networks has been repositioning itself as an integrated cybersecurity provider, with recent deals including a $3.35 billion purchase of cloud firm Chronosphere. Applied Compute’s funding round earlier this month valued it at $3.25 billion, with Nvidia among its backers.

Source: Palo Alto Networks’ announcement and Reuters’ report

Share to: , and

Faheem R. Khattak

Senior automation engineer and AI marketing journo covering LLMs, algorithms, analytics, core updates, Bing search, and technical SEO.

0 Comments

Leave a Reply

Optional Take Tag:

🛡️ Spam protection: Honeypot + Time-Trap Active

This site uses Akismet to reduce spam. Learn how your comment data is processed.