OpenAI Jalapeño: Better than Nvidia Blackwell

Post by dpe82 » Tue, Aug 25, 2026, 10:34 PM UTC

vntok wrote: post #76

dpe82 wrote: post #50

Earlier quoted context omitted.

NVidia has enormous operating margins, so a competitive solution doesn't have to match or beat NVidia's scale efficiencies; it just has to beat delivered cost. One objective of the project might be simply to provide credible negotiating leverage when dealing with existing suppliers like NVidia. You don't have to deploy at scale for that to work, but you do have to look like you could if pushed hard enough.

> NVidia has enormous operating margins, so a competitive solution doesn't have to match or beat NVidia's scale efficiencies; it just has to beat delivered cost. But then that means you have no actual moat against the behemot, right? Your competitor can move into the market as soon as they want to, at much better cost (so at slightly better price)... and Nvidia certainly can adapt much faster around hard hardware spe…

Post by Eridrus » Tue, Aug 25, 2026, 10:39 PM UTC

porridgeraisin wrote: post #118

Eridrus wrote: post #66

Earlier quoted context omitted.

Cerebras is targeting a distinctly different point on the cost/latency curve. They are betting that there will be some high value applications where latency and not just throughput is super important.

It is being used as part of a combined system. For example AWS is pushing for Trainium + WSE 3. The WSE 3 does the decode and the Trainium does the prefill. Even in nvidia land rubin + LPU does a similar thing. It has its downsides of course - if your traffic swings prefill heavy to decode heavy, you can't suddenly use your lpu for prefill. With GPUs they're totally interchangeable. Tradeoffs.

Post by derefr » Tue, Aug 25, 2026, 10:43 PM UTC

RealityVoid wrote: post #162

vineyardmike wrote: post #114

Earlier quoted context omitted.

How much of that 16mo is design versus just production? If there was a “plug and play” chip where you just BYO weights, how long would it take? The bigger issue seems to be that these chips can’t hold that many weights at the moment. (I’m curious if chips with large weights in them would be more tolerant or less to yield issues. If you flip a few bits in the weights, does it really matter at scale?)

Talaas, from what I understand is building stuff just like that. The infra is the same and the weights layer is all you need to change. I guess you could half etch the chips and then finish them with the weights only. I think their turnaround is 6-8 Weeks. The size of the models fitting on the chips at the moment is llama 3 I think?

Post by anukin » Tue, Aug 25, 2026, 10:55 PM UTC

htrp wrote: post #107

mchusma wrote: post #85

I think they talked about this being general purpose chip but I would think that Anthropic/OpenAI are at the scale now they could bake LLM weights into chips themselves. For example, GPT Sol baked into a custom chip run for $100M that runs 10x as fast and 10x as cheap should pay for itself as long as the chip is useful for long enough. While 2 years ago nothing was useful more than 1 year long, there are many older m…

etched tried this.... it didn't go very well

Post by subtlejellyfish » Tue, Aug 25, 2026, 10:56 PM UTC

jimmySixDOF wrote: post #11

I love how now you have to consider the possible s** posting motivation behind analysis of a trillion dollar industry being conducted at a world-class level by a bunch of ex Reddit and 4Chan adjacent mods -- it's one of the best stories in AI that SemiAnalysis is not cut from the same cloth as Gartner McKinsey et al