Palo Alto Networks to Acquire Portkey to Expand AI Security and Control Infrastructure
Every system looks smooth on the surface until you trace where the decisions actually move, that invisible layer where requests get routed, filtered, approved or killed is where the real game is played, and Portkey built right there in the part most people ignore until it breaks, while Palo Alto Networks is now stepping in to own that highway with intent.
On April 30, 2026, Palo Alto Networks announced its intent to acquire Portkey, the AI infrastructure company quietly handling trillions of tokens per month while the rest of the market argues about prompts like it’s open mic night, no price tag splashed across headlines because none is needed when the pipes matter more than the paint and negotiations happen where leverage speaks louder than press releases.
Massive congratulations to Rohit Agarwal and Ayush Garg, the co-founders who didn’t chase noise but engineered control as CEO and CTO, building in that zone where things either scale or snap and proving their system could hold under pressure without breaking when it mattered most.
Portkey operates as an AI Gateway and control plane, meaning every request, every model call, every agent decision flows through a single layer of routing, monitoring, governance, and cost control, the kind of infrastructure that rarely trends but quietly decides which companies survive once AI leaves the demo phase and enters production with real stakes attached.
By February 2026, they secured a $15M Series A led by Elevation Capital with Lightspeed participating, building on a company founded in 2023 that was already pushing over 500B tokens across 125M daily requests while managing more than $500K in AI spend each day across 24,000 organizations, numbers that signal not experimentation but infrastructure operating at meaningful scale.
Nikesh Arora is not collecting tools but assembling control, and with Prisma AIRS positioned as Palo Alto Networks’ AI security platform, the need extends beyond visibility into enforcement at runtime, making Portkey the checkpoint layer where intelligence is monitored, governed, and constrained when necessary regardless of how advanced the underlying models become.
Portkey’s path highlights a pattern founders tend to overlook, where winning doesn’t come from being the loudest but from being the most necessary, building for the phase after hype when enterprises begin asking harder questions around governance, latency, cost, and risk while others remain focused on surface-level innovation instead of foundational durability.
In AI, generating answers gets attention but controlling the flow of questions determines power, and Palo Alto Networks has positioned itself to sit directly in that control layer as this next phase unfolds.