5 takeaways on Palo Alto Networks’ $4.5bn acquisition deal

The cybersecurity giant is expanding into the market for training secure open weight AI models with its planned acquisition of Applied Compute, even as the company sees unprecedented growth from its AI security offerings

Image:
Nikesh Arora, CEO, Palo Alto Networks

Palo Alto Networks is expanding into the market for training secure open weight AI models with its planned $4.5bn acquisition of Applied Compute, in a move that could rapidly outpace the company’s existing massive opportunities in cybersecurity alone, Palo Alto Networks CEO Nikesh Arora said on Wednesday.

As the company disclosed its latest quarterly results, Arora pointed to a variety of signs that Palo Alto Networks is already at the centre of many of the largest technology shifts now underway driven by the arrival of ultra-powerful frontier AI models. Surging demand for AI security helped to drive the expectations-beating results for the quarter, the company reported.

The planned acquisition of Applied Compute will add another substantial set of opportunities for the company, meanwhile, including helping Palo Alto Networks to further “get in the token flow” and generate revenue via the usage of AI models, he said during the vendor’s quarterly call with analysts.

As for how large the opportunity to train and secure enterprise AI could ultimately become, it could “dwarf” what Palo Alto Networks is seeing from cybersecurity currently, Arora said.

“This could be bigger than what we do today,” he said.

Meanwhile, Arora also raised the possibility that the rush to patch vulnerabilities is contributing to a recent series of high-profile outages impacting US businesses and consumers.

What follows are five big takeaways from Palo Alto Networks’ announcements and Arora’s comments Wednesday.

Applied Compute: open weight AI acquisition deal

The $4.5bn all-stock deal to acquire Applied Compute, whose technology helps businesses train and secure open-weight AI models, is expected to close during the third quarter of Palo Alto Networks’ fiscal 2027, which ends 30 April 2027.

After setting out to train its own AI models, Palo Alto Networks evaluated nine providers—and ended up deciding to pursue the acquisition of the top provider, Applied Compute, Arora said.

Compared to Applied Compute, “no other team matched their technical pedigree in post-training open-weight models,” he said.

Crucially, incidents of rogue AI agents disclosed this year have shown how difficult it is to anticipate every way an AI agent might behave, once it had access to an enterprise application, Arora said.

“We came to the conclusion that as we train our models for enterprise applications, you will have to train a supervisory AI model at the same time. [You must] build AI that watches this AI,” he said.

“I cannot go in and intercept and understand somebody else's application—because I don't know how it's supposed to behave because I didn't train it.

“So imagine if you can train your application, and have the same team train something that supervises the application—I think that's the future of AI security.”

The ultimate conclusion was, “we have to get in the training business to be able to build a good supervision business. And that's our bet,” Arora said.

For customers, the appeal of open-weight models is that a business can customise a model using its own data and decide where to run it, he noted.

“Every enterprise sits on decades of proprietary data, internal playbooks and institutional judgment that cannot be exposed to public models,” Arora said.

“Until now, they have lacked a trusted partner to post-train, run and supervise open-weight models safely on their own.”

Applied Compute’s researchers will also work on Palo Alto Networks’ own security products, with Applied Compute co-founder and CEO Yash Patil set to become co-chief AI officer after the deal closes, the company said.

Applied Compute reached $100m in annualised revenue within 18 months of its founding, according to Palo Alto Networks CFO Dipak Golechha.

By Palo Alto Networks’ fiscal 2029, the company expects Applied Compute to contribute more than $1bn in annual recurring revenue (ARR) to its next-generation security business, he said.

Patching causing outages?

In the wake of the release of frontier AI models that are proficient at discovering software vulnerabilities, critical-severity flaws are being disclosed at nearly seven times last year’s monthly average, Arora said during the call Wednesday.

“These discoveries are happening at a speed and intensity which is hard for IT teams to cope with,” he said.

The massively increased volume of fixes that are resulting appears to be behind some of the recent outages that have impacted many real-world and digital services, Arora said, though he did not specify which outages may have had such a connection.

“We believe many of the major disruptions of the last few weeks, including outages across major enterprises and governments, were partially due to unintended consequences of patching and configuration changes—as IT teams simply cannot patch these vulnerabilities fast enough without compromising their production environments,” Arora said.

“The significant volume of patches will continue to be required for at least the next 18 months, carrying ongoing risks of collateral damage,” he said.

“We expect the collateral damage will decrease over time as customers get smarter about understanding exactly what to patch and when to patch.”

These dynamics are helping drive demand for Palo Alto Networks’ “virtual patching” service.

The service provides IT teams with a way to block attempts to exploit a vulnerability, while they prepare to apply the underlying software fix, the company has said.

AI driving growth

Palo Alto Networks’ AI security platform, Prisma AIRS, has now reached more than 1,000 customers and nearly $170m in ARR just five quarters after its launch, Arora said.

According to Arora, it is now officially the “fastest-growing product in the history of Palo Alto Networks.”

Meanwhile, revenue for Idira, the identity security business built on Palo Alto Networks’ $25bn acquisition of CyberArk earlier this year, also continues to receive a major boost from demand for protecting identities and permissions associated with AI agents, according to Arora.

Idira revenue climbed 23 per cent, year-on-year during the latest quarter, he said.

“The rise of AI agents introduces an entirely new class of non-human identities that must be secured in real time,” Arora said.

Looking ahead, Arora expects businesses to increase cybersecurity spending in coming years in order to keep up with AI-related security challenges, he noted.

“You’ll see cyber budgets rise, I’m sure, in the next two years—and you will also see a degree of consolidation, because the spray-and-pray strategy has not worked in cyber,” Arora said.

Getting ‘in the token flow’

At the same time, Palo Alto Networks sees a major way for its security business to grow with the amount of AI customers use, even if the number of employees buying seats remains relatively stable, Arora said.

He pointed to Prisma AIRS, which inspects the tokens that AI models process.

“The more tokens [that] are used in the enterprise, the more you have to have inspection tokens,” Arora said.

Over time, he estimated, a customer’s spending on Prisma AIRS could amount to two percent to five per cent of its spending on AI tokens.

Likewise, the planned acquisition of Applied Compute would give Palo Alto Networks another way into that spending, Arora said.

The expectation is that customers looking to train and run a model for a business application will also be willing to pay to secure it, he said.

Platformisation traction continues

During the first quarter of Palo Alto Networks’ fiscal 2027, ended 31 October, the company added roughly 170 net-new platformisation deals as customers consolidated more of their security tools with the vendor.

That exceeded the previous two fiscal first quarters combined, and “platformised customers now represent two-thirds of our next-generation security ARR,” Arora said.

Overall, it’s clear that “customers are making long-term architectural bets with us,” Arora said.

In total for the quarter, revenue rose 39 per cent year over year to reach $3.44 billion, surpassing the Wall Street analysts consensus estimate of $3.31 billion.

Non-GAAP net income, meanwhile, reached $1.05 per share during the quarter, beating estimates by 8 cents per share.

This article originally appeared on CRN UK sister websiteCRN US.