According to that link it's spent about 1/3 to 1/2 the time less than now since 2011.

Additionally it was much less in 2017 (but had a peak higher in 2018).

This is also the largest and sharpest rise since 2010.

It seems possible to me that they are overbuilding and over estimating demand increase.

Similar to how WorldCom did during the internet bubble.

AI isn't going anywhere, but that doesn't mean a lot of datacenters won't fail and get snatched up by companies that didn't (much like the WorldCom capacity).

Can they drop prices if the majority of their capacity is spoken for over the next five years?

I'm willing to bet dwll, Lenovo, apple, and Asus all have some of that capacity booked too.

Maybe even some retailers.

I'm still waiting for Linux to feel as good as Compiz on Ubuntu 7.04

It's not terrible now, but there was a long period of

AI definitely has some value and income.

Not as much as it's being hyped as, but significant. Anthropic for example will get nearly 50 billion in revenue this year.

And the value of people using AI smartly is definitely not zero. We're in the late 90s internet part of this. In 10 years there will be AI that we take for granted making some companies billions and a lot of failed companies. There's no reason to think otherwise.

It's also not going to replace lawyers and doctors, it will be part of how they do their job better much like the internet though.

I don't think this will kill Microsoft, Google, Meta, Amazon or Oracle. Their worwr case is having to dump their AI programs, take a stock hit, and issue some bonds. They all have enough revenue for the foreseeable future that they won't bankrupt.

I thought federal vehicles didn't have license plates. They're invisible to flock.

From a business sense it is somewhat unstable.

For example if any investments were made WRT the earlier tarrifs, I bet they don't look so hot now.

Or maybe you invested in a large off shore wind project, how did the stability of the US turn out?

And for more balanced take (Trump is more recent, and extreme, but not the only problem) congress as a whole is unwilling to work together and pass laws so we end up with executive orders and whiplash on policy.

For an investment that will take years to complete and pay off the US is becoming risky in some areas. The risk is due to the instability of policy and law.

Ideally yes.

But not necessary within a product version. Mayve this will force them to make sure the realease version really is release worthy.

I'd even argue that for some uses the (more) consistent behavior of not having models update could be a benefit.

They're claiming 10x Cerebras (a 25kw system) performance.

100x performance per watt.

Even if they're overestimating / best casing the situation there is value to be had here.

The current frontier models have use. This could be made for one and depending on hardware cost be useful for a year and replaced (Nvidia/AMD hardware are advancing fast enough to be obsolete pretty quick too). I can see a business built around a product based on an open weight model using these in servers.

This is for end user/cloud provider products. And if these can be developed fast enough and not cost too much, there's likely a market.

That assumes you have a phone. Most of the time I'm in the store is because my phone is lost or broken.

Last time I was there inhad a phone with a blank screen.

They made me do the transaction through the app guessing where to click on the screen.

They'll just say that their AI policy optimized for team players...