@Cobtribefan People have been saying 'something big' for months. Jensen showing up doesn't always mean instant fireworks. I'll believe the compute deals when I see the papers.
@josephcurl This is the part of AI progress that actually matters for most people. High-end models are cool, but shrinking the hardware requirements makes it usable for regular businesses.
AI isn't just getting more powerful.
It's also getting way more affordable.
TurboVec, based on Google Research's TurboQuant, can cut AI vector memory from 31GB down to 4GB.
That means:
Less memory needed.
Faster searching.
AI that works offline.
No need for pricey GPU setups.
No relying on the cloud.
Everyone's focused on bigger models.
But the real game-changer might be dropping costs.
If running AI gets cheaper, more people will use it.
Stocks to keep an eye on:
$GOOG $NVDA $MSFT $AMZN $META $AMD $AVGO
My take:
The next big AI trade isn't just about who builds the biggest model.
It's about who makes AI cheaper, faster, and simpler to roll out.