Palo Alto and NTT DATA chase $1B in AI security sales by 2029
Two enterprise moves today show where the money is flowing in AI: security vendors are betting that defending against AI-made threats means selling AI-made defense, and the autonomous-driving race is increasingly a silicon race.
Palo Alto Networks and NTT DATA signed a multiyear global alliance that targets $1 billion in joint AI-security business by the end of 2029. The deal, announced Thursday, is the first of its kind Palo Alto has struck with a global systems integrator, putting its security platforms in front of NTT DATA's consulting, engineering and managed-services customers with investment and forward-deployed engineers from both sides. The partnership isn't starting from zero — the two have run a joint managed-security service since 2024 and both sit in Palo Alto's Frontier AI Alliance — but the new agreement goes further, with a dedicated go-to-market model and a shared engineering roadmap that gives NTT DATA early access to features before they ship.
The centerpiece is the autonomous security operations center, which pairs agentic AI tooling with NTT DATA's managed detection and response staff — the idea being that AI handles the triage flood while humans handle the calls that matter. Identity security (built on Idira, the platform Palo Alto assembled from its $25 billion CyberArk acquisition), AI governance, multicloud posture management, firewall modernization and a combined zero-trust/secure-edge architecture round out the launch scope, with financial services, healthcare, manufacturing and the public sector first in line. Nikesh Arora framed it as operationalizing "platformization at true global scale," and NTT DATA's Abhijit Dubey argued customers need resilience built on AI-driven security and global reach. The $1 billion figure spans contracts and orders on both companies' books over three years, mixing product, subscription and services revenue — and notably the alliance is non-exclusive, so NTT DATA keeps reselling rivals like CrowdStrike. It's a clear bet that the fastest-growing line in security is no longer a product but a managed, AI-augmented service wrapped around a platform.
Waymo detailed the custom chip at the heart of its autonomous-driving stack, and it's purpose-built silicon rather than a GPU bolt-on. Alphabet's self-driving unit operates about 4,000 vehicles across 11 U.S. cities, and its sixth-generation compute module leans on two custom ASICs — chips designed from the ground up for one job — that together clear more than 1,000 trillion calculations per second. The accelerators are built on TSMC's 5-nanometer process in a vehicle-hardened edition that meets three automotive reliability standards, and Waymo runs two of them so the car keeps driving if one fails.
The chips are tuned for inference on sensor-fusion models that merge readings from 13 cameras, four lidar units and six radars into a single view of the road, with an error-correction step the company calls temporal denoising that checks adjacent frames to filter noise. Waymo's executives say the whole system runs entirely onboard, making driving decisions within milliseconds, and the module is ruggedized to survive constant vibration, shock and extreme temperatures while drawing cooling from the host vehicle's battery loop. The takeaway: as robotaxi fleets scale, the winners are designing their own silicon for the exact latency and redundancy a car demands — the general-purpose GPU is no longer the default answer.
Should enterprises hand routine security triage to agentic AI, or does that just widen the blast radius when the automation gets it wrong? Tell us in the comments.
