Thread by @wmhuo168

NVIDIA just bent the knee at the RISC-V Summit. CUDA now runs on the same architecture China is using to escape U.S. tech control. Here's why that should terrify Washington

RISC-V’s Ascent Could Reshape The Global Compute Landscape share.google/1uysrSEflxhmhY…

NVIDIA just joined the RISC-V Summit to announce CUDA support on RISC-V hosts. Let’s be clear. This isn't leadership. It's capitulation. (1/11)

For years, CUDA was NVIDIA's proprietary fortress. Arm and x86 were the only citizens. RISC-V was treated like a hobbyist toy. Now it’s getting top billing. Why? (2/11)

Because China made RISC-V real. Alibaba, SpacemiT, Huawei, and entire state policy turned it into a national escape hatch. This isn't just about chips. It's about compute independence. (3/11)

While the West debated benchmarks, China shipped billions. RISC-V cores now run AI chips, datacenter CPUs, embedded systems. The full software stack is already in place. (4/11)

The RISC-V Foundation moved to Switzerland in 2019. Not for alpine views. It was a shield against U.S. export laws. That’s what "open" really means in this context. (5/11)

Now NVIDIA shows up at the summit they once ignored. CUDA suddenly works on RISC-V hosts. Just like that, the walls came down. (6/11)

Not out of idealism. Out of fear. China is building an AI ecosystem that runs entirely without CUDA. NVIDIA needs RISC-V more than RISC-V needs NVIDIA. (7/11)

It's desperation in a tailored suit. If CUDA can't run in China, then China builds something else. NVIDIA knows that. They're trying to stay relevant. (8/11)

Notice they didn’t roll this out with European or American vendors. They’re chasing the Chinese market that Washington is actively trying to cut off. (9/11)

This is how the West loses control over compute. Not through sabotage. Through arrogance. We laughed while China industrialized the ISA. (10/11)

RISC-V is not a science fair. It’s the blueprint of a multipolar tech world. And the West keeps acting like it’s still 2012. (11/11)