Catch up on the essentials
- Intel has officially confirmed that its upcoming Razor Lake-AX processor is targeted for launch in 2028 with approximately 700 trillion operations per second (TOPS) of AI computing power aimed at robotics and humanoid machines.
- The company plans to bring Nova Lake to market for robotics in early 2027, with Razor Lake-AX following in 2028.
- Intel has not specified what numerical precision the number is based on, nor how much of the performance gain will come from the CPU, GPU, and neural processing hardware.
Selected from this article · 2026-10-17
Read on for the full pictureIntel outlines Razor Lake-AX as a robotics-first platform

Intel has officially confirmed that its upcoming Razor Lake-AX processor is targeted for launch in 2028 with approximately 700 trillion operations per second (TOPS) of AI computing power aimed at robotics and humanoid machines. The confirmation came from Intel Senior Vice President Cho Min-sung during a presentation of the company's roadmap at the Semiconductors Powering Humanoids conference in South Korea on Oct. 16, as reported by The Elec.
The announcement frames Razor Lake-AX as a possible competitor to AMD's Ryzen AI Max processors, while explicitly positioning the chip at on-device inference for robots that must perceive and decide without round-trips to remote servers. Intel has not confirmed whether the processor will also arrive in consumer PCs.
700-TOPS target framed against Panther Lake's 180 TOPS
Intel's current robotics platform based on Panther Lake offers up to 180 TOPS of combined AI performance, putting the stated 700-TOPS Razor Lake-AX target at roughly 3.88 times that figure. The company plans to bring Nova Lake to market for robotics in early 2027, with Razor Lake-AX following in 2028. Intel forecasts mass production of autonomous mobile robots for some customers by 2027, and humanoids by 2028, aligning with the chip's reported timeline.
For humanoid robots that must coordinate perception, movement, and real-time control, on-device processing can help reduce latency and dependence on cloud connectivity. That additional processing power could enable robots to analyze camera feeds, identify objects, understand their environment, and make decisions without continuous server links.
Specs, precision and power remain unspecified
The 700-TOPS figure is described as a roadmap target rather than a finalized performance spec. Intel has not specified what numerical precision the number is based on, nor how much of the performance gain will come from the CPU, GPU, and neural processing hardware. Power consumption, memory capacity, and real-world performance are also listed as unknown in the coverage, and the figure should not be taken as proof that robots running on the chip will be four times faster than current Panther Lake systems.
More information on the design, power requirements, and final specs for Razor Lake-AX will be needed before the 700-TOPS target can be assessed against competing platforms.
Other developments
Separately, coverage also describes Razor Lake-AX as the first commercial SoC from Intel to combine a wide memory bus with large integrated graphics silicon, integrating either 16 or 32 Xe3P graphics cores while retaining standard Nova Lake core architecture, targeting real-time machine vision and AI inference for autonomous robotics hardware.
Share this article






