Ed Zitron draws conclusions from Microsoft's downscale of future data centers. Ed interprets this as MS losing faith in the future of AI but it could be a turn away from brute-force model size expansion and toward DeepSeek-style efficiency improvements.
I've been waiting for "Lottery Ticket Hypothesis" breakthroughs and I hope this is the beginning.
