Big Tech's $1.1 Trillion Cloud Computing Backlog (sherwood.news) 17
An anonymous reader shares a report: Amazon, Google, and Microsoft each reported hundreds of billions in RPO (remaining performance obligations) -- signed contracts for cloud computing services that can't yet be filled and haven't yet hit the books. Collectively, the big three cloud providers reported a $1.1 trillion backlog of revenue.
The push for data centers (Score:3)
I guess this is the source for the push for the new data centers. I can sort of understand the reason the accounts like it. From a developer standpoint, outside of production where you want regional redundancy, I don't understand wanting to push so much work to "someone else's computer" that you don't have full control over it and where you can make the accountants and VPs pull there hair out because you ran a development process a few extra times and now they have a computer bill they weren't expecting, but that's what it took to get the job done. Also, putting less in the data center means you spread the electrical demands around the country instead of forcing it all in 1 area to the detriment of the people in that area.
neo-mining (Score:5, Informative)
Re: (Score:3, Interesting)
Our county in Indiana just rejected two data centers, and while the exact concerns were different (power costs, water usage) the sentiment was the same: the big cloud companies will not stripe-mine our resources.
Backlog of revenue? (Score:3)
So the companies that owe that trillion dollars are a trillion dollars in debt? What happens when this trillion dollar backlog just doesn't happen because the companies that booked it disappear because they can't service their trillion dollar debt? Sounds like it's time to head for the door...
Re:Backlog of revenue? (Score:4, Interesting)
Nothing happens, except huge stock market drops.
As far as the companies themselves. Its all funny money. Company A promises to use 200 Billion in compute from hyperscaler B over Y years if hyperscaler B invests $X, where $X is slightly less than 200 Billion over Y years, paid in installments contingent on something that almost certainly requires hyperscaler B to do the build out for A to do the contingency work.
So basically its all flim flam use to to impress^H^H^H^H^H^H bilk investors, where by these guys got to go out on stage and talk about the millions or billions in deals they have with other tech corp from the news, but the reality is the AI boys don't have applications for their product and the hyperscaler boys don't have chips to run it on; so a whole lot less activity is actually happening than appears.
Its all a big cow-milking game, sucker larger investment mangers into buy buy buy buy! While the startup capital diversifies themselves out of way before anyone catches wind there is way less real growth going on.
Re: Backlog of revenue? (Score:1)
Backlog (Score:5, Informative)
I can't speak to all of the areas that contribute to AI backlog (like capital allocation, systems integration, networking availability, etc). But from a data centre standpoint it is a real struggle. The general timeline to get from requirements to signature on a data centre lease is about three months if all goes well (assuming you are not self-performing). Once that is signed a DC project takes about two years from conception to RFS (ready-for-service) where you can start rolling in racks. A LOT goes into that timeframe: Ordering LLE (Long-Lead Equipment (generators have a 24+ month lead time right now)), securing permits, securing power, securing a EOR (Engineer of Record) and getting the design done, land preparation, construction crews and materials, etc.
And that is all just to get a "powered shell": what we call a building with at least two diverse sources of utility power, gens, transformers, power switch gear, static UPSs (if they are being used), chillers, pumps and piping, etc. You still can't roll in racks until you have a fit-out design (PDUs, RPPs, whips, busway, FWUs (Fan Wall Units) or CRAHs (Computer Room Air Handlers), temp and humidity sensors, etc. It takes about a YEAR from the time you get the fit-out design to the time when you can roll in racks (but it is possible to line up fit-out works with RFS if you get the design early enough).
Everybody is looking for some creative solutions to speed things up, but data centre deployments are complex (I deal with well over 1000 individual requirements per project) and there are a lot of supply chain constraints. Most of the hyperscalers are juggling 20+ simultaneous projects, it's burning people out and there is a lot of churn in the business (making it worse).
Re: (Score:2)
And let's not forget that when they're asking for gigwatts of power they now have to wait for the power plants to be built, which are a little longer lead time than backup generators...
Things are bad enough the brokers are contacting every single hole-in-the-wall with a closet calling itself a colocation facility if they have space for a dozen or two racks with 100kW power requirements. That is, 100kW *per rack*.
Imagine what our economy would be like (Score:1, Troll)
At a certain point you have to start asking if the structure of our civilization benefits you or not.
If you're over 65 you're probably going to die before the damage is done but if you're under 65 you either need to start asking that question or you need to start picking out yo
ELI5 here... overbooked data centers? (Score:2)
If I'm reading this right, pretty much the companies are overbooked, with not enough data centers to handle the incoming workloads? Or is this meaning that their clients are in arrears?
I have been worried about a "cloud cliff". In previous recessions, businesses could weather bad times by dropping service plans, not updating (when updates were paid), or even going on the high seas and surviving that route. Now, with cloud subscriptions, if a company can't afford to make their cloud bill, there is no redu
\o/ (Score:2)
I wonder if they'll deliver before the bubble pops
Re: (Score:2)
They will deliver some the data centers and hardware and then those big 3 and a few others will be stuck making payments on the bonds for hardware and data centers that they may not have any significant customers and they hopefully are profitable enough that they won't go out of business getting stuck holding the bag.
The people making the purchases seem be purchasing the capacity because they think the magical AI is going to make them lots of money, if it does not (and that seems pretty likely that it is go
Re: (Score:2)
Re: (Score:2)
Unlikely. This is not something that can be done fast. I wonder who will not survive the crash.
Re: \o/ (Score:1)
Everyone? It's gonna be big