Compute Trade and Silicon Information, Bochev added “are additionally engaged on growing clearer benchmarks for the compute market, and may have extra particulars to share on that within the coming weeks.”
PIC ‘ought to serve to maintain suppliers trustworthy ..’
Scott Bickley, an advisory fellow at Information-Tech Analysis Group, stated he views the providing “as a method for enterprises to supply short-term GPU capability and probably get a deal, particularly whether it is stranded capability from the neocloud suppliers.”
This, he stated, “would additionally assist to benchmark prices when buying this capability usually, so it’s good, however additionally it is easy by way of the worth proposition.” He additionally famous that the majority corporations are usually not shopping for GPU capability straight; “That is for these which can be constructing their very own fashions or deploying their very own AI functions atop current fashions.”
Bickley added, “it ought to serve to maintain suppliers trustworthy to some extent by way of the flooring and ceilings of the worth to entry GPU capability.”
Quickly after Compute Trade first launched in February, Matt Kimball, VP and principal analyst for information middle compute and storage at Moor Insights & Technique, described the GPU compute scenario as “fairly dire. That is pushed by what most view as a single provider (Nvidia) promoting GPUs earlier than they will even be made to a market that has an insatiable thirst.”
On Tuesday, following the announcement, he stated that the idea of PIC is interesting: “I actually like the thought of PIC as a instrument for purchasers and seeing the compute trade change into an arbitrageur of types. This delivers an actual worth to [anyone] who’s trying to make the most of AI infrastructure,” he stated.
