who cares whether it is the fastest for AI workloads? there are plenty of other workloads. this machine is fastest at HPL benchmarks - and that represents important applications. frankly, i /am/ interested it is made with home grown cpus and interconnects.

I think I would be pretty dismayed to hop in my car to head out for work in the morning and discover that it dumped half it's charge for datacenters... and that GM took profit out of that, to boot!

I would suggest that if reduced staff can handle the current backlog, then the current staff could do even more. Why isn't the backlog "fattening"? Because the sales organization already worked their asses off to develop the current backlog. The company does not have the expertise or know how to extend to new markets; the economy is hard; competition is fierce - it is not easy. But the fundamental reason to downsize is you are not able to develop a productive workload to make the workforce profitable. In that case, you've got to downsize to maintain a profit margin.

AI coding tools are tools. Leaders give their people tools. You don't let your people go because you give them good tools. The point where you let your people go is the same as always: because for one reason or another, you failed to find enough work for them to do.

what developer would want to work for a company that has so many developers that it can lay off 8000 and still function? holy cow, how dehumanizing is the sheer size alone? i could not imagine it.

"Amazon had already surpassed both carriers to become the nation's largest parcel shipper by volume..." Not correctly written. The author is comparing apple and oranges there. A carrier transports items while a shipper hires a carrier to transport items.

Of course, it is really THE MIXING that should be done by robot, 'cause it's the daubbing part people would enjoy doing with their own hands!

yes, they can be designed do specific tasks better than people - ever hear of the industrial revolution??

Soul of a New Machine was practically my anthem when I was young upstart. Also loved House, which I found many years later. Though I am on the computer side, I recognized myself in both stories. ("The screwdriver is used up" is from Hackers by Steven Levy about early computer days at MIT.)

Pardon me, but isn't code review before merge to production normal practice? If senior engineers aren't reviewing, then I would say that the problem is not the use of AI tools...

Let us just clarify: LLM tools i(ncluding ChatGPT) do not.produce "answers" - they ptoduce "responses"!

Oh wait, that's Kashi...

So rather than distribute compute to us locally, where we can use it as we wish, pay once, and even buy it second hand, we're going to shift users to cloud services where use is restricted to specific apps (games, LLM APIs), users pay forever (different rates for different apps), and those jerks justify their resource hungry over-built data centers!

Yes to version control with LLM tools! FYI, Aider already has it integrated. In fact, use version control for everything you do on the computer. (Then backup your repo, need I say it?)

"This validates the 'do more with less' strategy..." I would say, "supports," not "validates." I use LLM tools every day to support my code development. It is not so easy to ensure good enginering while using an LLM based tool. While I feel that I am faster with the tools, I doubt it is really possible to produce quality output wrangling 5 sessions on 5 pieces of system code at a time.