The AI Sift is part of you-do-nothing

← Back to The Latest

Palo Alto Networks buys Applied Compute for $4.5 billion, betting every firm runs open-weight AI

Palo Alto Networks agreed to buy Applied Compute for $4.5 billion in stock on Nov. 18 to help firms train and run open-weight models in a safe way.

The decision it puts on your desk

Map where you train or run open models by Nov. 26. List each place, who keeps it safe, and where the data lives. If one place uses a firm that is not your security firm, get a price from Palo Alto and one other firm before your next deal. Set your rules before Palo Alto closes the deal in mid 2027.

Palo Alto Networks said Nov. 18 it will buy Applied Compute for $4.5 billion in stock to sell safe open-weight AI to firms.

Palo Alto Networks hero image (Source: Palo Alto Networks)
Palo Alto Networks hero image (Source: Palo Alto Networks)

Transaction terms and platform scope

Palo Alto signed a deal to buy Applied Compute. The firm builds a platform for post-training and inference of open-weight models.

Applied Compute's Applied Compute Agent Cloud (AC2) is the core system. Its team uses it to train open models, run them widely, and feed live feedback into the next version. Customers named by the firm include Microsoft, Nvidia, Cognition, Mercor, DoorDash, and Harvey.

"Every enterprise is going to run a mix of open and closed models, and every one of those models needs to be secured," said Nikesh Arora, chairman and chief executive of Palo Alto Networks. "Applied Compute has built the best post-training and inference platform we have seen, and its research team is exceptional. We saw that firsthand as a prospective customer, and our Unit 42 Frontier AI Assessment confirmed their strong security foundations."

Yash Patil, co-founder and chief executive of Applied Compute, will become Co-Chief AI Officer of Palo Alto Networks after close. Patil said the tie-up puts safety at the core of the platform. The firm will keep serving AI-native firms, digital natives, and large firms. Palo Alto will take the platform to its global base.

The price is $4.5 billion in stock, subject to change. Close is set for Palo Alto Networks' financial third quarter of 2027, waiting for approval from regulators. Palo Alto will share more on its financial first quarter 2027 call at 1:30 p.m. Pacific Time (PT) on Nov. 18.

Ownership map showing why a security firm holds a model platform
Ownership map showing why a security firm holds a model platform

Strategic rationale for security platform expansion

Palo Alto frames the bet as hybrid. The firm said the future for firms is a mix of open and closed models tuned to firm data, cost, and control needs. No large provider sells those open models in a safe way today. With Applied Compute it plans to be that partner, in public cloud or in home-country sites.

That bet adds a new sales line to a security sale. Palo Alto sells network, cloud, security work, and identity. AC2 gives it a training and run layer that firms can host where rules require. The pitch covers training, tuning, and running open-weight models in a safe way.

Shares moved on the news. Palo Alto raised its yearly sales forecast with the deal. Coverage noted the stock rose about 8 percent on earnings.

Buyer adoption costs and integration risk

For buyers the key is where the model runs and who keeps it safe. AC2 helps teams tune open models on firm data. It also runs them and uses feedback to make them better, per coverage of the platform. A security firm that owns that loop makes proof of source, access control, and safe hosting a single deal rather than two.

For Palo Alto the cost is merge risk. The firm must keep Applied Compute's research speed while joining it to a 70,000-customer security platform. Stock as deal cash helps the books. The platform must now serve two groups at once, AI-native builders and large firms that buy through Palo Alto.