Brainchip's AKD1500 Edge AI hardware now sells for $99 and up. The lineup includes M.2 and PCIe accelerator cards plus the BrainBoard 1500 SPI module. All three products have moved from announced to purchasable, according to CNX Software.

The entry price matters. At $99, the cards undercut most discrete AI accelerators aimed at embedded developers. Whether that price holds at volume is another question. Brainchip has not published volume pricing in the announcement.

What does the AKD1500 family include?

The release covers three form factors:

• AKD1500 M.2 card — an accelerator in the M.2 module standard (a small plug-in card format common in laptops and compact industrial PCs)
• AKD1500 PCIe card — a desktop and server form factor using the PCIe expansion slot standard
• BrainBoard 1500 SPI module — a development board that connects over SPI (Serial Peripheral Interface, a short-distance chip-to-chip bus)

CNX Software lists availability "for $99 and up." The floor price likely corresponds to the smallest module, not the PCIe card. The seller has not published a per-SKU price table in the source material.

Who is the hardware for?

Brainchip builds its products around a neuromorphic processing approach, a chip design that processes events rather than full data frames and can run at low power. That positions the AKD1500 family at edge sensing applications — vision, audio, vibration detection — where power budgets sit in the milliwatt range rather than watts.

The SPI module targets developers who want to pair the accelerator with a host microcontroller. The M.2 and PCIe cards target integrators who need a standard socket. One form factor per integration path, and nothing more.

What is not said?

The announcement omits several details a buyer would need. There is no stated shipment date for preorders versus stock on hand. There is no performance benchmark against comparable accelerators at the same price. There is no stated software stack version.

Buyers should verify three things before committing: which SKU actually carries the $99 price, what toolchain support ships with it, and what the return terms are for evaluation units. The source headline confirms availability and a price floor, nothing else.

At $99, the cost of finding out is low. That appears to be the point.