Jensen Huang compared Bittensor’s distributed AI training milestone to “our modern version of Folding@home,” while TAO jumped from $243.5 to $310.6 before easing, according to the report.
Nvidia CEO Jensen Huang discussed Bittensor on the All-In podcast after Chamath Palihapitiya pointed to a recent Subnet 3 training run as a major technical accomplishment. Huang responded by calling it “our modern version of Folding@home,” framing the project’s distributed AI training effort in familiar terms for mainstream tech audiences. The report also said a correction was later posted stating the run involved a 72 billion-parameter model, not 4 billion, and was trained permissionlessly across more than 70 contributors. Following the remarks, TAO rose from $243.5 to $310.6 before cooling to about $297-$298 by press time. Huang did not endorse the token directly, but said proprietary and open models should coexist and argued that specialized industries will need controllable open models.