Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Qualcomm Moderated Thread - please read rules before posting -- Ignore unavailable to you. Want to Upgrade?

To: QCOM_HYPE_TRAIN who wrote (195896)	10/6/2025 10:43:38 AM
From: Wildbiftek	4 Recommendations Recommended By kech Lance Bredvold sbfm waitwatchwander Read Replies (2) \| Respond to of 196679

I've always thought that GPUs had a lot of cruft like texture mapping units and ROPs that are a waste of die area and power when it came to dedicated neural-net MAC and activation calculations. TCO contains a big piece related to data center operations and the better density and efficiency of dedicated solutions would make them the longer term winners. This does seem like AMD doing a pivot from the traditional laptop client space where Qualcomm is starting to embarrass them into data centers where during this frothy time of reckless cap-ex, metrics like efficiency may matter less than speed of deployment or upfront costs. It's unclear how politics or software stacks are affecting this, but ARM locking horns with Qualcomm seems to have temporarily hobbled its deployment in both client and datacenter solutions. I still think Qualcomm has a lot to offer in its higher efficiency in both TOPs and CPU performance as well as its best in class integration of the cellular subsystem. In the longer run, getting inferencing results to clients and collecting telemetry for further reinforcement training will require what exactly Qualcomm has to offer, but I also think it's desirable to saturate clients with this style of compute to run more focused versions of models for privacy and latency / and relegate data center models to "thinking" versions that augment on device results.