Kochi: A startup founded by three former OpenAI researchers has landed a $4.5-billion acquisition deal, less than two years after its founding.

Palo Alto Networks has entered into a definitive agreement to acquire Applied Compute, an artificial intelligence company specialising in training and deploying open-weight AI models for businesses.

The all-stock transaction is valued at $4.5 billion, subject to adjustments, according to the cybersecurity company's announcement on November 18. The acquisition is expected to close in the third quarter of its fiscal year 2027, subject to regulatory clearance and other customary conditions.

Following the completion of the deal, Applied Compute co-founder and CEO Yash Patil will become Co-Chief AI Officer at Palo Alto Networks. The startup's research team will also join the cybersecurity company.

Who is Yash Patil?

Yash Patil founded Applied Compute alongside Rhythm Garg and Linden Li. The three founders previously worked as researchers at OpenAI, contributing to different areas of its artificial intelligence development.

The founders established Applied Compute around the idea that businesses should be able to develop and operate AI models tailored to their own requirements, rather than depending entirely on general-purpose models.

Patil reiterated this vision while announcing the acquisition on X. He said the company was founded on the belief that every business should own its intelligence, with the ability to train, serve and improve proprietary AI models.

He also highlighted the company's growth, saying a team of just a few dozen people had taken it from zero revenue to more than $100 million in annualised revenue run-rate in just over 18 months.

Why is Palo Alto Networks acquiring an AI startup?

The deal reflects Palo Alto Networks' strategy to expand its role in enterprise artificial intelligence while addressing the security challenges associated with its adoption.

In its November 18 announcement, the company said it expects businesses to use a combination of open-weight and closed AI models, with models increasingly customised to their data, costs and control needs.

Open-weight models allow users to access a model's released parameters and adapt it for specific applications. This gives businesses greater flexibility and control over how they deploy AI.

However, enterprises also need to ensure that these models can securely access sensitive information and systems, comply with internal policies and operate reliably.

Applied Compute's platform is designed to support the post-training and inference stages of AI development. This includes adapting models for specific business requirements and running them in production environments.

Palo Alto Networks plans to combine these capabilities with its cybersecurity infrastructure, allowing enterprises to train, customise and deploy AI models across public cloud and sovereign environments.

Palo Alto Networks chairman and CEO Nikesh Arora said the company wanted to help businesses securely use both open and closed AI models.

He also said Applied Compute's research team would help integrate frontier AI capabilities across the company's security products.

In his statement on X, Arora said the new team would play a key role in making Palo Alto Networks an AI-first company.

For Patil, the acquisition is an opportunity to scale the company's original mission with security at the centre of its products. He said customers would continue to own and control their models.

The proposed acquisition brings together two areas increasingly converging in enterprise technology: artificial intelligence and cybersecurity. However, the transaction is yet to receive the necessary regulatory clearances and be completed.