Incident sheet · /enterprise-ai

Filed
Read
4 min · 779 words
Severity
Routine

Palo Alto Networks Pays $4.5B in Stock for Applied Compute

Palo Alto Networks will buy Applied Compute for $4.5B in stock to train and host open-weight AI models, targeting 100MW of compute and a new co-chief AI officer.

Palo Alto Networks bets $4.5B on enterprise AI with Applied Compute acquisition
FeedPalo Alto Networks bets $4.5B on enterprise AI with Applied Compute acquisition — SuzyLu / pixabay

Event log

  1. Palo Alto Networks agreed to buy Applied Compute for US$4.5 billion in an all-stock deal announced with fiscal Q1 2027 results.
  2. The deal is expected to close in Palo Alto's fiscal Q3 2027, running February to April; Applied Compute CEO Yash Patil becomes co-chief AI officer.
  3. CEO Nikesh Arora said open-weight models grew from under 5% to over 60% of token traffic in a year and targeted 100+ MW of compute in 3-5 years.
  4. The acquisition follows Palo Alto's US$25 billion purchase of CyberArk.
  5. Palo Alto has sold model-agnostic protection via Prisma AIRS since 2025.

Palo Alto Networks has agreed to buy Applied Compute for US$4.5 billion in an all-stock deal, pushing the security vendor directly into training, deploying and hosting open-weight AI models for enterprises in the public cloud or their own sovereign environments.

The company announced the acquisition alongside its fiscal first-quarter 2027 results, calling Applied Compute "a leading platform for post-training and inference of open-weight AI models." The deal is expected to close in Palo Alto's fiscal third quarter of 2027, which runs from February to April. Applied Compute co-founder and CEO Yash Patil will become Palo Alto's co-chief AI officer, and the startup's research team will build AI models across the company's product line.

"The acquisition will bring Applied Compute's platform and world-class AI research team to Palo Alto Networks to help enterprises adopt AI securely, whether they choose open-weight models, closed frontier models, or both," the company said in its statement. Applied Compute will keep serving AI-native companies, digital natives and large enterprises after the deal closes.

Why does a firewall vendor want to run models?

Chairman and CEO Nikesh Arora told analysts on the earnings call that beyond embedding AI in Palo Alto's own security products, "we will help train and serve secure sovereign implementations of open-weight models to the enterprise." He claimed open-weight models have grown from under 5 per cent of token traffic to more than 60 per cent over the past year, though he did not cite the source of the figure.

Pressed by JP Morgan analyst Brian Essex on customer demand, Arora argued that deterministic security controls "cannot anticipate every scenario that an agent is going to conduct." Enterprises building AI applications "will have to train a supervisory AI model at the same time," he said, because Palo Alto cannot secure models it has no hand in building. "I cannot go in and intercept and understand somebody else's application," Arora said.

Sanchit Vir Gogia, chief analyst at Greyhound Research, sees the logic but flags a gap. A generic filter can't know how a specific application should behave — Arora has a point there. But most AI application failures happen at runtime, through prompts, tools and permissions, regardless of who trained the model. Palo Alto has already sold model-agnostic protection through Prisma AIRS since 2025.

"A supervisor trained by the same hands as the system it supervises is a colleague, not a check," Gogia said. His advice: CIOs should ensure someone other than the trainer tests the result.

GPUs, data centres and the sovereignty question

To train and host those models, Palo Alto will build its own compute capacity. CFO Dipak Golechha said on the call the spending "will mostly go to GPUs and servers," with a smaller share for the facilities housing them. The company will "continue to source from our cloud partners" and add dedicated capacity over time. Arora set the target at more than 100 megawatts of compute within three to five years.

Sovereignty is where the plan gets thin, per Gogia. Palo Alto has credentials for security data — customer-held keys and a European Cortex service governed by Deutsche Telekom — but "none of that yet extends to GPUs." A US-headquartered operator faces the same jurisdictional questions as the hyperscalers it rents capacity from.

"A sovereign hosting offer with no named region, operator or date is a promise looking for an address," Gogia said. Compliance officers should demand a region list, the operator and a dated certification plan.

Arora, for his part, told analysts the training and inference business could eventually exceed Palo Alto's core market: "This could be bigger than what we do today."

What should Applied Compute customers do before closing?

The deal follows Palo Alto's US$25 billion purchase of CyberArk, which added identity security. Applied Compute now adds the models themselves to a vendor already running many customers' firewalls, SOC and identity controls.

"Palo Alto is asking enterprises to let the guard build the vault it guards," Gogia said. One team watching both model and telemetry can spot trouble sooner — but a single fault, breach or pricing dispute would land "on the protection and the protected together."

His checklist for Applied Compute customers before the deal closes:

  • Secure ownership of trained weights in an exportable format
  • Bar training on their data without consent
  • Win a right to exit on change of control
  • Get price protection that survives bundling
  • Put a costed exit in writing before any model work moves across

"Terms are cheapest to win while the suitor is still courting," Gogia said.

via arnnet.com.au (Source)

Filed under

  • palo-alto-networks
  • applied-compute
  • enterprise-ai
  • open-weight-models
  • ai-infrastructure
Share this article:

More from Olivia Hart

Olivia Hart

Show full bio

News editor covering money, funding and market shifts around enterprise information technology and the leaders who run at CIO Monitor.

215 articles

The newsroom digest

One email a day with the stories that matter. No noise.

By subscribing you agree to the privacy policy.

Cross-references

‹ Previous articleNext article ›