DiffuSpace raises several hundred million yuan in two rounds, largest dLLM deal

The signal here is less the amount than the investor mix: a dLLM specialist pulling in both traditional VC and strategic backers tied to Huawei and Horizon suggests diffusion-based language modeling is being evaluated against real deployment pipelines, not just research benchmarks. That matters for the broader AI stack, where dLLM's pitch is faster, parallel generation rather than the token-by-token approach dominant today. Whether this concentration of capital translates into production-grade adoption, or remains a well-funded research niche, is the open question worth tracking.