Palo Alto Networks agreed on Wednesday to buy Applied Compute, an 18-month-old startup that helps companies train and run open-weight AI models, in an all-stock deal valued at $4.5 billion. The announcement landed alongside a record quarter for the cybersecurity giant — and it signals a bigger bet: that the next phase of enterprise AI will be won not by whoever builds the best model, but by whoever makes open models safe enough for large companies to actually use.

The deal is striking for its speed and its price. Applied Compute was founded in 2025 by three former OpenAI researchers, raised $350 million at a $3.25 billion valuation earlier this month, and is now being bought for roughly 40% more than that — before its second birthday.

What Palo Alto is buying#

Under the definitive agreement, Palo Alto Networks will acquire Applied Compute in an all-stock transaction valued at $4.5 billion, subject to adjustments. The deal is expected to close in Palo Alto Networks’ fiscal third quarter of 2027, subject to customary closing conditions including regulatory clearance.

Applied Compute’s platform handles post-training and inference for open-weight models — the kind whose trained weights are published for anyone to download, adapt and run on their own infrastructure. Its pitch is to make it practical for a company to take one of those models, tune it on data it would never send to an outside provider, and run it in production, with an autoscaling inference engine and a model router called Shift-1 doing the heavy lifting.

The people come with the platform. Co-founder and chief executive Yash Patil will become Palo Alto Networks’ Co-Chief AI Officer once the deal closes, and the startup’s research team will join the company. Palo Alto Networks says Applied Compute will keep serving its existing customers — AI-native companies, digital natives and large enterprises — while the platform is rolled out to Palo Alto Networks’ own global customer base of more than 70,000 organizations.

An engineer at a workstation studying a large screen showing a neural network with a glowing padlock at its center
AI-generated illustration: Applied Compute’s platform adapts open-weight models to a customer’s own data — with security controls built into each step.

A startup that grew up fast#

Applied Compute was founded in 2025 by Patil, Rhythm Garg and Linden Li, all former OpenAI researchers. Its rise has been unusually compressed even by AI-industry standards: earlier this month it announced a $350 million funding round at a $3.25 billion valuation, led by Elad Gil with Nvidia, Shopify Ventures and HubSpot Ventures participating, and said its annualized revenue run rate had topped $100 million. Notable customers include Microsoft and DoorDash, and the company had raised more than $500 million in total before the acquisition.

That trajectory explains the price. Palo Alto Networks is paying a premium over a valuation set only weeks ago, for a company whose revenue is still a rounding error next to its own — but whose team and product sit exactly where the enterprise AI market is heading.

The bet: hybrid AI needs a security layer#

Palo Alto Networks’ argument is that enterprise AI is going hybrid. A large share of companies, it says, will run open-weight models tuned to their own data, cost and control needs alongside closed frontier models — and today no large provider is delivering those open models securely. With Applied Compute, enterprises will be able to train, adapt and deploy open-weight models with security built in, in the public cloud or in sovereign environments built to satisfy compliance and data-residency rules.

“Every enterprise is going to run a mix of open and closed models, and every one of those models needs to be secured,” Arora said in the announcement. The company also gets a model lab in the purchase: Applied Compute’s researchers will help train Palo Alto Networks’ own proprietary large and small models and embed AI across every one of its products.

The deal continues a buying spree: Palo Alto Networks has spent 2026 pushing beyond its traditional cybersecurity business, including the identity-security company CyberArk and the observability firm Chronosphere. Applied Compute is the most pointed move yet — it puts the company into building and tuning models, not just defending them.

Analysts silhouetted before a wall of monitors showing world maps and threat-detection dashboards, one screen showing an open box merging into a shield
AI-generated illustration: the pitch is that open and closed models alike will need securing — and that security, not the model, is the product.

The earnings backdrop#

The acquisition was announced on the same day Palo Alto Networks reported what Arora called a record first quarter. For the quarter ended October 31, the company reported revenue of $3.44 billion, up 39% from a year earlier and ahead of the $3.31 billion analysts expected, with adjusted earnings of $1.05 a share against 97 cents expected. Shares rose more than 8% in late trading, adding to a gain of nearly 10% during the regular session.

The company also raised its full-year guidance materially: revenue is now forecast at $14.52 billion to $14.57 billion, up from $14.10 billion to $14.20 billion guided in September. A strong quarter gives Palo Alto Networks room to pay up for a strategic asset — and the market’s reaction suggests investors liked both halves of the day.

What to watch#

Three questions will decide whether this looks smart in hindsight. First, the close: the deal needs regulatory clearance and is not expected to complete until Palo Alto Networks’ fiscal third quarter of 2027. Second, the integration: Palo Alto Networks has to turn a research-heavy startup into a product its 70,000 customers buy, without breaking the team and the platform that made it worth $4.5 billion. Third, the category itself: “secure open-weight AI” has to become a real product line, not just a pitch. If it does, a cybersecurity vendor becomes an AI infrastructure vendor — and the $4.5 billion will look like an entry ticket, not a splurge.

Sources: Palo Alto Networks, “Palo Alto Networks to Acquire Applied Compute, Bringing Secure Open-Weight AI to the Enterprise”; SiliconANGLE, “Palo Alto Networks bets $4.5B on AI startup Applied Compute as earnings top forecasts”; Unite.AI, “Palo Alto Networks Reaches $4.5B Agreement for Applied Compute”.