Nvidia adds AI peanut butter to Nokia's 6G network chocolate, throws in $1 billion

Nvidia CEO Jensen Huang on Tuesday announced a partnership with Nokia to integrate AI technology into its mobile network infrastructure, bringing accelerated computing to the edge and paving the way for 6G-ready networks. As part of the deal, Nvidia will invest $1 billion in Nokia. Team Green's gear will boost spectral …

[ecofeco]

Oh this will end well

Not.

[williamyf]

if one can not run the inference device side (for varying definitions of device), and one can not wait for the data to arrive to a hyperscaler data center to run the inference, I guess the shelter at the bottom of the antenna mast* is kind of a nice place to run said inference.

*or failing that, the same datacenter where the regional 6G core resides

[Kevin McMurtrie]

Re-failing the lessons of 5G

• Higher bandwidth efficiency is good and this is the only reason anyone upgraded to 5G
• Nobody cars about ultra-fast high bands with unusable range
• 5G resulted in lots of neighborhood small cells, none of which have edge compute
• Radio physics still exists and uninterrupted mobile data still doesn't exist
• Most of the 5G specifications will never be used

And here's 6G to build upon everything that was a failure in 5G. Absolutely no telco is going to use edge compute yet. Possibly some day, but not in the lifespan of 5G or 6G.

[DS999]

Re: Re-failing the lessons of 5G

Higher bandwidth efficiency is good and this is the only reason anyone upgraded to 5G

Not true, the 5G most of us use doesn't have higher bandwidth efficiency than LTE.

The main reason 5G is faster is that there were a lot of new bands added for it, or recovered from other uses like 2G or TV. The latest 5G standard allows chopping up the bands into small pieces which helps reduce congestion when busy (similar to wifi 7 in that respect) but that's not widely deployed yet (you get it as part of "5G standalone") 5G also allows grabbing bigger chunks of spectrum at once so you can get really high speeds when you have little contention where you are, that's why the peak 5G speeds without contention are a lot better than the peak LTE speeds without contention, but those wide channels don't come into play when you're on a busy cell.

The places where I get really fast 5G is what Verizon calls "5G UW" which isn't mmwave but 3.5 and 6 (ish) GHz bands. If all those bands had been added as LTE then LTE would have been a lot faster. I'll get over 500 Mbps consistently on those if it isn't too busy, which is faster than there's really any need for at least for a phone. There's zero benefit to ever upgrading to phones to 6G, doubly so if it builds AI into the network core! If they want to add bands there's no reason they can't add them for 5G and improve the experience there.

[druck]

AI circle jerk

Sounds like another AI circle jerk investment.