
RISC-V MCU Adds Built-In Ethernet PHY and 8MB PSRAM
WCH's CH32V467 packs a 200 MHz RISC-V core, built-in Fast Ethernet PHY, 480 Mbps USB, and up to 8MB PSRAM at roughly $2.40 to $2.90 per chip.
Integration Is the Whole Story
WCH has built a reputation on RISC-V microcontrollers that undercut expectations on price. The CH32V407 and CH32V467, detailed on August 13, 2026, extend that with a specification that removes two of the most annoying external components in embedded networking designs.
- 32-bit RISC-V core at up to 200 MHz, with RVV vector extensions and 4.11 CoreMark/MHz
- 10/100 Mbps Ethernet MAC with integrated PHY, eliminating an external PHY chip
- Up to two USB 2.0 high-speed interfaces at 480 Mbps with built-in PHY
- CH32V467 adds 4MB or 8MB on-chip PSRAM; both parts carry 200KB SRAM and 512KB flash
Chips run roughly $2.40 to $2.90 in bulk through AliExpress, with development boards at $17.99 to $21.99 from AnalogLamb.
Why Does an Integrated Ethernet PHY Matter?
Anyone who has designed a networked microcontroller board knows the tax. The MCU gives you a MAC, and then you add a separate physical-layer chip, magnetics, and the careful impedance-matched routing between them. That external PHY costs board space, adds a part to the bill of materials, and introduces a layout problem that is easy to get subtly wrong.
Putting the PHY on-die removes all of that. A two-dollar part becomes a complete Ethernet endpoint with the addition of a magjack. For industrial sensors, building automation nodes, and the kind of networked hobby projects that would otherwise reach for a much more expensive board, that changes what is economical to build.
The 480 Mbps USB high-speed PHY is the same argument in a second domain. Full-speed USB at 12 Mbps is common at this price; genuine high-speed USB with an integrated PHY is not.
What Can You Do With 8MB of On-Chip PSRAM?
The CH32V467's PSRAM is the specification that separates it from an ordinary networked MCU. Microcontrollers in this class typically offer a few hundred kilobytes of RAM, which caps what you can hold in memory at any moment.
Eight megabytes changes the category of problem you can attack. It is enough for a framebuffer, and the chip includes an LCD-TFT display controller and a DVP camera interface to make use of it. A part that can drive a display, capture from a camera, hold a frame in memory, and reach the network over built-in Ethernet is doing the job of a small applications processor at microcontroller cost and microcontroller power.
The peripheral list is deep for the price: up to 77 I/O pins mappable to 16 external interrupts, 3 SPI, 10 USART, CAN 2.0B, two I2S audio channels, dual 12-bit DAC and dual 12-bit ADC, in LQFP64, QFN68, and LQFP100 packages.
Where This Fits in the RISC-V Landscape
WCH continues to be the company making the strongest practical case for RISC-V at the low end, where architecture arguments matter less than cost per capability. We recently covered the $6 CH32V203-based Comu board that hides inside a USB port, and the CH32V407/467 sit a clear tier above it in ambition.
For makers and industrial designers working through our single-board computer and embedded coverage, the practical takeaway is that networked, display-capable embedded projects just got cheaper to design. Not because a headline benchmark moved, but because two chips and a routing headache came off the board.
Sources: CNX Software — August 13, 2026; WCH product documentation — August 2026.
More Mini Computers Stories

MaTouch ESP32-S3 E Ink Board: A $60 Voice-Ready Dashboard
Makerfabs' MaTouch ESP32-S3 pairs a 7.5-inch four-color E Ink screen with a four-mic array for $59.80. Here are the specs and what you can build.

ASRock NUC 300 Mini PCs: Wildcat Lake With Dual 2.5GbE
ASRock Industrial's NUC 300 series brings Intel Core 5 320 and Core 3 304 to mini PCs and boards with up to 64GB of DDR5-6400 memory and dual 2.5GbE.

DGX Spark 64GB: Who NVIDIA's $4,999 Local AI Box Is For
NVIDIA's DGX Spark 64GB starts at $4,999 on Oct 23 and runs models up to 100B parameters. Here is who the local AI box suits and how clustering works.
