AI will trigger 'hyper-scale demand' of data centers, says Morgan Stanley's Laurel Durkay
HTML-код
- Опубликовано: 20 авг 2024
- Laurel Durkay, head of global listed real assets at Morgan Stanley, joins 'The Exchange' to discuss the potential demand for data centers driven by AI, companies positioned to face the demand, and more.
This is OK but Roaring Kitty tweeting about GME is somehow wrong? Give me a break
sort of misleading. if AI-producing firms are competing and compressing inference costs for a given set of outcomes, then the accelerator hardware needed for a certain volume of outcomes needs only so much climate control. in fact, one may be able to serve a certain volume of outcomes with consumer hardware and residential property w/o minimal management. so the "underlying need for infrastructure" is really just a demand for whichever nearly-commoditized arrangement can bargain between so many independent providers.
there is a critically erroneous overestimation of ML scaling because of the current transformer/diffusion hype cycle.
Just like the early computers took up the size of a school gym.
nah, you're wrong
Data Center Technician checking in. 💪 In regards to AI, the systems that run AI, can draw as much power as a single house. This is also depends on how many GPU'S are in a single server. Typically a standard Server rack is 48U with 16-24 servers. GPU beefed up AI systems reduce the amount of space based on sheer size and power draw. You would be lucky to get 8 AI systems in a single rack, due to power restrictions. These systems can blow exhaust heat around 130 degrees Fahrenheit. The cooling to hot air containment along with electricity are the limiting factors.
You cannot compare consumer grade computing to business grade servers and systems. Even the older legacy servers can pull 2k watts or more, depending on what's installed and running. There is no way you can run a 96 core or higher CPU, which is used for AI on limited power or in a personal setting.
Nuclear power
The second of two new nuclear reactors in Georgia has entered commercial operation, capping a project that cost billions more and took years longer than originally projected.
$CEG Constellation Energy as they are the largest nuclear operator? Or Duke Energy as I believe they are close 2nd in nuclear? Who will rule them all or will it be a diversified game. Heck, I think there are about 4 nuclear operators; but Constellation Energy certainly operating the most.
The amount of time to get a nuclear power plant up and running can take decades. AI is not going to wait that long. Our grid is not ready, period.
@@BryantMitchell We have nuclear plants all over the US and world. As AI optimizes and advancements in storage and all metrics, Nuclear will lead us until Fusion. Likely diversified renewables and nuclear until we reach fusion.
@@BryantMitchell and yes, new plants can take decades to deploy. emerging technologies like SMR are very promising. Bill Gates started an energy company- TerraPower. Also, NuScale Power amongst others are actively testing and designing R&D.
I love the grounded reality of this channel!!,Despite the recession, I no longer depend on Government Grants since I acquire $16,400 bi weekly profits.....
I've been talking to Andras Bohm Consultant, for long now, mostly because I lack the knowledge and energy to deal with these ongoing market circumstances. there are more aspects of the market than the average individual is aware of. Having an investing counselor is now the best line of action, especially for those who are close to retiring..
Run a quick online research with his name
his website should appear right away..
Scam