← Back to Blog
RISC-V MCU

Haawking HX2000: The RISC-V DSP Line Behind a Fresh Series B — And What Still Isn't Published

2026-09-07 6 min read EN
RISC-V MCU

Table of Contents

The round

Beijing Zhongke Haoxin Technology (中科昊芯) announced on 2026-09-03 that it has closed a Series B round, with Yuehu Capital (悦湖资本) as the investor. Coverage followed on 2026-09-04. The amount was not disclosed.

The company said proceeds go specifically to AI edge-side chip product R&D and to industry-customised solution development — not to general working capital. That distinction matters, because it tells you where the product line is heading.

Company background, as published

What the HX2000 series is, per published material

The Haawking HX2000 series is the company's in-house RISC-V DSP line, and is stated to be in volume deployment. Named application areas:

The edge-AI variant is described as built on a self-developed RISC-V DSP plus dual-core heterogeneous architecture, targeting ultra-low power, on-device AI inference, real-time compute response and high stability. Named target applications are robot joint motion, vehicle powertrain, and smart appliance deployment.

That is the whole of what is published about the architecture. There is no clock frequency, no DMIPS or MMACS figure, no memory configuration, no peripheral list, no package or pin count, and no power number in any source retrieved for this draft.

Why RISC-V DSP is a distinct engineering bet

DSP is a segment where the RISC-V argument is unusually clean, and it has nothing to do with licences:

  1. Deterministic control is the product. Motor control, digital power and PV/storage inversion are real-time loops. What matters is interrupt latency, deterministic instruction timing, and the quality of the PWM/ADC/capture peripherals — not a Dhrystone number.
  2. Toolchain ownership is the moat. A DSP line lives or dies on its compiler, debugger and library quality. A vendor that owns its toolchain end to end can tune it against its own pipeline. Zhongke Haoxin claims exactly that in-house chain.
  3. Long product lifecycles. Industrial and automotive programmes run ten years or more. An ISA with multiple implementers is a different continuity proposition than a single-vendor core.

The counter-argument is real too: a proprietary DSP toolchain is a lock-in of its own, and "in-house toolchain" tells you nothing about code quality or standards conformance without a benchmark.

Segment context

Two data points give the round its context:

Read together: capital is moving into RISC-V control-plane silicon, not only into RISC-V application processors.

What is not disclosed

None of the following appears in any source retrieved. Nothing is estimated here:

If you are evaluating this line, the questions above are the ones to put to the vendor in writing. The published material will not answer them.

Sources

Verification notes

Open RISC-V — RISC-V Silicon & Development Board Store — open-riscv.com