Dev.to AI πŸ€– Ai πŸ‘ 0 πŸ“– 6 min read

Palo Alto Networks to Buy Applied Compute for $4.5 Billion in All Stock Deal

Palo Alto Networks moved deeper into the enterprise artificial intelligence stack on November 18, 2026, signing a definitive agreement to acquire Applied Compute, a San Francisco startup that post-trains and serves open-

Palo Alto Networks to Buy Applied Compute for $4.5 Billion in All Stock Deal

Palo Alto Networks to Buy Applied Compute for $4.5 Billion in All Stock Deal

Palo Alto Networks moved deeper into the enterprise artificial intelligence stack on November 18, 2026, signing a definitive agreement to acquire Applied Compute, a San Francisco startup that post-trains and serves open-weight AI models, in an all-stock deal valued at $4.5 billion. Announced alongside fiscal first quarter 2027 results, it would fold a young model-tuning specialist into a vendor best known for firewalls, cloud security and its Unit 42 threat research arm. Palo Alto Networks said the deal should close in its fiscal third quarter of 2027, subject to customary conditions including regulatory clearance.

The logic on offer is a hybrid one. Palo Alto Networks argued in its announcement that a large share of enterprises will run on open-source and open-weight models tuned to their own data, cost and control needs, and that today no large provider is delivering those models securely. Applied Compute's platform adapts existing open-weight models to a customer's own data and then runs them in production, in public clouds or sovereign environments. That, the company contends, lets it sell security around the models themselves, not only around the applications and networks around them.

The purchase is not isolated. Palo Alto Networks bought Console, an AI-native agentic workflow platform, in September 2026, and chief executive Nikesh Arora has been positioning the company as a broader security provider as businesses adopt AI. This deal extends that arc from protecting AI deployments to helping build them. Applied Compute will keep serving AI-native companies, digital natives and large enterprises, while Palo Alto Networks brings the platform to its global customer base.

Applied Compute arrives with unusual momentum. Acquiry Deal Intelligence reported on November 19, 2026, that the company had stated a $100 million annualized revenue run rate reached less than 18 months after founding, alongside more than 1 trillion tokens per day of inference volume on its AC Inference service. On November 2, 2026, it announced $350 million in new funding at a $3.25 billion valuation, with investors including Nvidia, Blackstone, Kleiner Perkins and Lux Capital. Against that price, the $4.5 billion consideration represents a premium of roughly 38 percent in 16 days.

Key Facts

The mechanics come from the merger documents. Acquiry Deal Intelligence reported on November 19, 2026, that the $4.5 billion total, subject to adjustments, is paid entirely in Palo Alto Networks common stock at an issue price of $337.16 per share, the volume-weighted average closing price on Nasdaq for the 15 trading days ending November 17, 2026. The company had about 821 million shares outstanding as of November 11, 2026, and its Form 8-K states that the number of shares to be issued is not yet known.

Closing conditions include expiration or termination of the Hart-Scott-Rodino waiting period, with a close expected in fiscal third quarter 2027. Applied Compute's research team joins after close, and co-founder and chief executive Yash Patil becomes Co-Chief AI Officer. SecurityWeek reported on November 18, 2026, that the research team will also help train proprietary models used across Palo Alto products, a detail that matters as much as the platform itself.

Arora framed the deal around model mix rather than model loyalty. "Every enterprise is going to run a mix of open and closed models, and every one of those models needs to be secured," he said, adding that Applied Compute has built the best post-training and inference platform the company has seen and that its research team is exceptional. Palo Alto Networks had seen the platform firsthand as a prospective customer, and its Unit 42 Frontier AI Assessment confirmed Applied Compute's strong security foundations. Patil said the combination will put security at the center of all its products.

The deal landed next to a strong quarter. Calcalistech reported on November 18, 2026, that revenue for the quarter ended October 31 rose 39 percent year over year to $3.44 billion, ahead of guidance of $3.30 billion to $3.31 billion, while next-generation security ARR rose 66 percent to $9.73 billion against a guided $9.54 billion to $9.56 billion. GAAP net income was $482 million, or $0.57 per diluted share, versus $334 million and $0.47 a year earlier.

Profitability was mixed in the details: GAAP operating income fell to $96 million from $309 million, while non-GAAP operating income rose to $1.1 billion from $746 million, and cash from operations was $1.9 billion, a 56.2 percent adjusted free cash flow margin. Guidance moved up. Full-year fiscal 2027 revenue is now $14.52 billion to $14.57 billion, up 26 percent to 27 percent and raised from $14.1 billion to $14.2 billion, with next-generation security ARR of $11.475 billion to $11.525 billion and non-GAAP earnings of $4.26 to $4.30 per share. Chief financial officer Dipak Golechha said the company is materially raising its full year guidance.

Analysis

The obvious read is that Palo Alto Networks is paying up for a hot asset. The bigger picture here is that security vendors are being forced to decide whether AI models are something they guard from the outside or something they ship, tune and operate for customers. Palo Alto evaluated Applied Compute first as a prospective buyer and had Unit 42 assess the platform before acquiring the supplier. Testing a product as a customer and then buying the company behind it shows where Palo Alto believes the budget is going.

What this really means is that the open versus closed debate inside large enterprises is being settled as a portfolio question rather than a doctrinal one. Arora's formulation, that every enterprise will run a mix of open and closed models and that every one of those models needs securing, is the whole thesis in a sentence. If that is right, the durable position is not owning the best model but owning the tuning, deployment and policy layer around many models. A $100 million run rate reached in under 18 months suggests demand for that layer is real.

The price is also a statement about scarcity. At $4.5 billion for a stated $100 million run rate, Palo Alto Networks is paying roughly 45 times annualized revenue, a growth multiple rather than a value multiple. That only makes sense if the company expects to pull the platform through an installed base far larger than the startup could reach alone. The all-stock structure softens the accounting sting, and the $337.16 issue price ties the real cost to Palo Alto's own share performance.

There are reasons for caution. Post-training and inference is a crowded field, and Applied Compute's own backers, among them Nvidia, Blackstone, Kleiner Perkins and Lux Capital, had marked it at $3.25 billion only on November 2, 2026. Paying about 38 percent above a two-week-old valuation is a bet on speed, on getting the team before a rival does. The deal also depends on regulatory clearance under the Hart-Scott-Rodino process, and because the share count is not fixed, the final consideration will drift with the stock before close.

Why It Matters

For Palo Alto Networks customers, the practical change is that model customization and model security would arrive from one vendor. Enterprises that have been stitching together open-weight models, tuning pipelines and guardrails from separate suppliers could buy post-training, inference and the controls around them together, deployable in public cloud or sovereign environments. Sovereign deployment is the part that matters for regulated buyers barred from placing data in a shared hyperscale tenant.

For the security market, the deal widens the category. The core business is growing quickly without it: revenue up 39 percent to $3.44 billion and next-generation security ARR up 66 percent to $9.73 billion in the first quarter. That strength funds an adjacency bet and raises the bar for rivals. A competitor that only secures AI workloads at the network edge now faces a vendor that also supplies the model-serving layer underneath them.

For investors, the verdict was immediate. Calcalistech reported on November 18, 2026, that the stock had already gained nearly 10 percent in the regular session ahead of the earnings release and climbed another 8 percent after hours. The market read the acquisition and the raised outlook as one signal, and the guidance behind that reading is hard to argue with.

Next Up

The near-term milestones are procedural. Closing is expected in fiscal third quarter 2027 and hinges on the Hart-Scott-Rodino waiting period expiring or terminating, along with other customary conditions. Until then Applied Compute keeps serving existing customers, and investors will ask whether the $100 million run rate holds. Palo Alto Networks promised further detail on its earnings call held on November 18, 2026 at 1:30 p.m. PT.

The harder test comes after close, when Patil takes the Co-Chief AI Officer role and the research team is asked to do double duty: run a commercial platform and help train the proprietary models used across Palo Alto products. If that works, $4.5 billion buys a second engine inside a security company. If it does not, it buys an expensive tuning shop in a market where the underlying models keep getting better and cheaper.

πŸ“° Read the original article on Dev.to AI

Originally published by Dev.to AI. Aggregated on AIWithGhost for educational purposes β€” full credit and traffic to the original publisher.