A security vendor buys a model platform
The stated logic is a bet on how enterprises will run AI. Palo Alto Networks' chief executive, Nikesh Arora, puts it in the release: "Every enterprise is going to run a mix of open and closed models, and every one of those models needs to be secured." Applied Compute is the platform it is buying to act on that view: a platform for post-training and inference of open-weight AI models, whose research team the release says will also help train Palo Alto Networks' own proprietary large and small models and embed AI across its offerings.
That makes this a different kind of acquisition from the security tools Palo Alto Networks usually buys. It is not adding a detection feature to an existing console. It is buying the layer where open-weight models are adapted and served, plus the research team behind it. If the deal closes as planned, a firewall vendor will also be in the model-platform business.
The price, read against the target's own numbers
Applied Compute gave the market fresh numbers about itself only weeks before this deal. Its November 2 announcement reported a $100 million annualized revenue run rate, alongside $350 million in new funding at a $3.25 billion valuation (figures via Unite.AI's account of the announcement). The acquisition values the same company at $4.5 billion, subject to adjustments. That is 45 times the reported run rate, and about a 38% step up from a private valuation set earlier the same month.
Two qualifications belong next to those multiples. The run rate is the company's own figure, reported in its own funding announcement, and a run rate annualizes a recent period rather than recording a year of revenue. And the consideration is stock, not cash: Palo Alto Networks shareholders are paying with dilution, and the $4.5 billion is a value at announcement, subject to adjustments, on a deal that is not expected to close until fiscal 2027.
What has to happen for $4.5 billion to make sense
Earnings-day coverage put specific targets on the plan. Security Boulevard's summary of the earnings-day discussion reports management projections of more than 100 megawatts of inference capacity within three to five years, capital spending of less than $3 billion between fiscal 2027 and 2030, and $1 billion of added annual recurring revenue by fiscal 2029. None of those figures is in the acquisition press release, all of them are projections, and the coverage does not pin down whether each applies to Applied Compute alone or to the combined company's broader AI plans. The revenue figure is the one the price will be judged against, and it is also the one with no reported quarter behind it yet.
The regulatory path also remains open. The companies expect to close in the fiscal third quarter of 2027, which leaves months for clearance. The open-weight ecosystem Applied Compute serves does not stand still while a deal waits to close.
Sources
- [1] Palo Alto Networks, "Palo Alto Networks to Acquire Applied Compute, Bringing Secure Open-Weight AI to the Enterprise," PR Newswire, November 18, 2026 — read first-hand: the definitive agreement, the $4.5 billion all-stock value subject to adjustments, the fiscal Q3 2027 expected close, the Patil appointment.Read source
- [2] Unite.AI, "Palo Alto Networks Reaches $4.5B Agreement for Applied Compute," November 18, 2026 — Applied Compute's November 2 run-rate and funding figures.Read source
- [3] Security Boulevard, "Palo Alto Networks To Acquire Startup Applied Compute for $4.5B," November 18, 2026 — the employee count and the inference-capacity, capital-spending and ARR targets from earnings-day coverage.Read source