Intel unveils Neural Compute Stick 2 for developing AI algorithms and computer vision systems locally using its Movidius Myriad X AI chip, for $100
Ahead of its first AI developers conference in Beijing, Intel has announced it's making the process of imparting intelligence into smart home gadgets …
Context & Ripple Effects
This is the third step in a deliberate ladder Intel has been climbing since its 2016 push into AI silicon, when it laid out a Nervana-centered AI chip strategy that was explicitly incomplete until the Movidius acquisition closed. Movidius then shipped a $79 USB Compute Stick on the 12-core Myriad 2, and last August Intel launched the Myriad X VPU itself.
The Neural Compute Stick 2 is where those threads meet: the new VPU repackaged into the same thumb-drive form factor at $100, announced ahead of Intel's first AI developers conference in Beijing — a signal that this is a developer-ecosystem play, not just a component drop.
First-order effects
- Developers building smart-home gadgets and computer vision systems can now prototype against the Myriad X locally for $100, instead of waiting for OEM devices that embed the chip.
- Intel gains a direct channel to put Myriad X silicon into makers' hands ahead of its Beijing conference, seeding the software ecosystem around the VPU it launched in August.
Second-order effects
- Nvidia's answer came within months: the Jetson Nano board at $99 targets the same entry-level developers and makers, turning sub-$100 AI dev hardware into a head-to-head Intel-Nvidia pricing and ecosystem contest.
- Cheap host-side acceleration shifts value toward whoever owns the toolchain — each vendor's stick is only as sticky as its SDK, so software support becomes the differentiator rather than the silicon.
Third-order effects
- If the pattern holds, edge AI development hardware commoditizes at the ~$100 price point, pressuring both companies to compete on integrated stacks — chips plus frameworks plus deployment paths — rather than standalone parts.
- Local inference on commodity USB hardware normalizes running vision models on-device, pulling AI workloads out of the cloud and into smart-home and sensor products by default.
The trend: AI compute is cascading from datacenter silicon down to sub-$100 developer hardware, with Intel and Nvidia racing to lock makers into their respective edge ecosystems.