Intel officially confirmed Razor Lake-AX, a high-performance AI processor intended for robotics and humanoid machines, with a target of 700 TOPS by 2028.
The company positions Razor Lake-AX as a potential competitor to AMD’s Ryzen AI Max and as a way to push on-device AI processing for real-time robot control.
Intel plans Nova Lake for robotics earlier, in 2027, with Razor Lake-AX following in 2028, but exact specs and power figures remain unconfirmed.
Real-world details on precision, power use, memory, and how the mix of CPU/GPU/NP components achieve 700 TOPS are still uncertain, so treat the 700 TOPS as a roadmap target rather than a final spec.
Quick read · 1 min
Intel unveiled Razor Lake-AX, a robotics-focused AI chip aiming for 700 TOPS by 2028. It would be a big step up from Panther Lake and could push more on-device AI in robots, with Nova Lake planned for 2027.
What this means for you: faster, more autonomous robots and devices that can handle perception and control locally, without always talking to the cloud. Exact specs and price aren’t known yet, and Intel hasn’t announced consumer devices.
What’s next: more details on performance, power, and availability should come closer to 2027–2028 as Intel rolls out Nova Lake first and Razor Lake-AX later.
Intel is signaling a bold step into on-device AI for robots. At a robotics-focused industry gathering in South Korea, the company publicly outlined Razor Lake-AX, a processor it says could deliver about 700 trillion operations per second (TOPS) for AI workloads in robotics and humanoid machines. The numbers are framed as a roadmap target for 2028, not a confirmed spec sheet yet.
The 700 TOPS goal would put Razor Lake-AX roughly four times faster in AI tasks than Intel’s Panther Lake robotics platform, which tops out around 180 TOPS. Intel also mentioned a separate future push called Nova Lake for robotics, aiming for early 2027. That gives a sense of how the company is mapping a ladder of chips meant to handle perception, decision making and control directly on the device rather than relying on cloud power alone.
What this means for everyday tech users is a future where robotics and advanced machines could work more independently. Fewer delays and less dependence on a constant internet connection could translate to faster object recognition, better reaction times in dynamic environments, and more capable on-board decision making. However, concrete specs on precision, memory, power consumption, and how much of the 700 TOPS comes from CPU vs GPU vs neural processing components aren’t official yet.
Intel’s push into robotics isn’t happening in a vacuum. The Razor Lake-AX plan sits alongside tools the company already markets for developers, like OpenVINO and Robotics AI Suite, and it could influence how robots are built and tested across factories, service robots, and research labs. For now, the company emphasizes the broader aim: give robots more autonomy with more capable on-device AI compute, reducing latency and cloud reliance.
01
Which robots would benefit first
Intel hasn’t pinned down exact product targets, but the roadmap suggests autonomous mobile robots and humanoids as primary beneficiaries. The emphasis is on real-time perception and control, not just smarter on-paper benchmarks.
02
What 700 TOPS actually means in practice
Top-line numbers are useful, but the real question is how much of that power comes with what kind of accuracy and energy use. Intel didn’t specify precision, power envelopes, or how the workload splits across CPU, GPU, and dedicated neural processors. So the 700 TOPS should be treated as a roadmap target rather than a guaranteed capability at launch.
03
What this could cost you (as a consumer or business user)
Right now, there are no consumer Razor Lake-AX devices or pricing to quote. The impact on everyday devices will hinge on whether Intel includes these chips in robotics platforms used in industry, home robots, or specialized devices. The company’s robotics toolkit, including OpenVINO, could help developers build apps that leverage Razor Lake-AX once it ships in mass production.
04
What happens next
Intel expects Nova Lake to show up earlier in 2027 as a robotics-focused platform, with Razor Lake-AX following in 2028. Precise specs, power targets, and partner plans will likely surface closer to those dates.
An AI lab will subsidize plumbing, electrical, HVAC, air-duct cleaning, and carpet cleaning in three cities in exchange for recording the work inside homes.
