technologyScore 55/100Watch
AMD Halo AI PC launches Q2 2025; claims 1.7x tokens/sec/dollar vs NVIDIA DGX Spark
“Ryzen AI Max delivers comparable performance to at much lower price than Nvidia's DGX Spark. Generating up to 1.7 times more tokens per second per dollar when running the latest GPT OSS models... Halo is launching in the second quarter of this year.”
Why it matters
AMD is directly positioning its Ryzen AI Max unified-memory silicon (128GB shared CPU/GPU/NPU memory) as a price-performance alternative to NVIDIA at the edge/local inference tier, with a concrete launch quarter.
Investment implication
Bullish read for AMD's client/AI edge franchise and a competitive datapoint against NVIDIA's DGX Spark ramp; suggests price pressure in local inference hardware could compress NVIDIA's edge AI TAM.