The agreement adds a plan for open-weight AI. Customers still need the terms of the service they use today.
Palo Alto Networks announced a definitive agreement to acquire Applied Compute on November 18. The company’s announcement values the transaction at $4.5 billion in stock, subject to adjustments. Regulatory clearance and other closing conditions remain. The companies expect closing in Palo Alto Networks’ fiscal third quarter of 2027.
What the agreement commits to
Applied Compute provides infrastructure to adapt and serve AI models. Palo Alto Networks says the proposed combination would help enterprises train, adapt, and deploy open-weight models across public and sovereign cloud environments. It also says the business will continue serving AI-native companies and larger enterprises.
The November 18 Form 8-K records the merger agreement and stock consideration. Its reference price for the buyer’s shares is $337.16, based on the stated averaging period. The filing says the final number of shares to be issued is still unknown. The announced transaction value therefore reflects a different quantity than a finalized share count.
Turn integration plans into procurement questions
| If you need this | Evidence to request from the service you would buy |
|---|---|
| Model adaptation on company data | The training workflow, supported models, and data-use terms for the selected product. |
| A particular deployment location | The service’s actual region, hosting arrangement, and retention terms. |
| Security integration | Which controls are available now and which are committed for future delivery. |
| Continued access after the deal | Contract terms for support, migration, pricing changes, and service termination. |
The checklist follows from the announced plans. For example, a company promising sovereign deployment gives a buyer a reason to ask for the exact region and operating terms. It leaves the buyer with a concrete document to assess before using sensitive training data.
Separate the agreement date from the service change
A founder already using Applied Compute can inventory the API endpoints, model artifacts, and credentials their application depends on. Keep the current commercial terms and note any later notices about changes. A new buyer can use the same inventory to specify what a pilot must demonstrate.
For that pilot, choose one supported model and a small authorized dataset. Compare the adapted model with the starting model on a held-out task under the same acceptance rules. Record the cost and latency alongside the result. That is a proposed procurement test. The announced transaction establishes the companies’ intended combination, while an actual service evaluation establishes what the customer can use.



