Yann LeCun says the company is leaving "no stone unturned" in chip effort, and that existing chips are largely inadequate for deep learning.
AI models are best trained right where the data is, and most customer data isn’t in IBM’s cloud.
From bare metal cloud to network fabric, partners will serve whatever needs edge computing users have on Vapor sites.
The company's researchers expect to improve AI compute performance by 1,000 times over the next 10 years.
