We're announcing a multi-generation collaboration with Amazon Web Services to build next-generation AI data center infrastructure, working together on customized silicon and AI inference at scale. Read the announcement: Read the 8-K:

@awscloud Custom silicon + inference at scale. AWS just opened a second lane.

@awscloud Congrats to the @Qualcomm team. Inference at that scale on your own silicon is a big one. And a good thing to be building next to.

Thank you. 🤝

@awscloud Custom silicon plus optical interconnect for inference at scale is the right bet - the bottleneck moved from training FLOPs to cost per token in production. A multi-generation AWS commitment is what gives that roadmap the volume to matter.

@awscloud Custom inference silicon succeeds when compiler support, model coverage, networking and rack operations mature with the chip. Multi-generation alignment makes that systems work possible.

@awscloud More hyperscaler deals please 😤

@awscloud Exactly. AWS stamp turns QCOM into real DC infra, not just phones.

@awscloud

@awscloud announcement diluting the shareholders.. seriously its time for @cristianoamon and @AkashPalkhiwala both of you bozos to resign. You guys run this Co down when its actually needed to boost it up. Always late. We've had it enuff..

@awscloud cool!

@awscloud Snapdragon Smooth Motion UI is a revolutionary technology. Like MediaTek's Dimensity Scheduling Engine 2nd Generation. Will the Snapdragon 8 Elite Extreme Gen6 (SM8975) introduce Snapdragon Smooth Motion UI？

@awscloud Custom silicon and inference scale are becoming the real battleground for data center efficiency.

@awscloud As AI moves into more enterprise workflows, scalable inference will matter just as much as model performance. Interesting direction for the industry.

Custom inference silicon still needs a floor poured under it. Data halls were long designed around 150 psf live load, and high-density AI floors are now being specified at 250 to 400 psf. If you're bidding an AWS-region shell, get that live-load number in writing before you freeze the slab this week.

@awscloud Inference silicon is where AI economics becomes real. The product users see is only the visible tip of a much deeper compute stack.