How would a national government (not TLAs) target particular individuals in a large number of users and what information can they gather given e.g. https://mullvad.net/en/help/no-logging-data-policy ? So perhaps not quite as easily as ordering a tap.

My national government has no business knowing which protocols I use to contact which endpoints and tamper with that traffic. Wrapping up that information in a tunnel is a good first protection layer.

You forget that nation-states control your ISP. And of course you can choose your VPN provider or run your own.

The provider and national TLAs will see all traffic that is in cleartext and meta traffic which is even more valuable. It can also actively tamper with that traffic. So you’re technically incorrect and you assume your threat model is universal. It’s not. And, of course, there are use cases for Tor, whether with or without VPN.

A free running cellular automaton (CA) approach in hardware would work, but each cell would be a much souped up SRAM cell, the interactions would be all local and 2D. Considering Cerebras is 40 G SRAM on the 300 mm WSI and is about at the cooling limit I’m afraid you do not have 5 orders of magnitude. Perhaps reversible spintronics can help with the power draw, but you still have to splat a higher dimensional network so not just local interactions into a 2D array.

No, that captures just the neuroanatomy. Not the properties like density of ion channels, type, value of the synapse and all the things we don’t know yet.

You seem to trust Nvidia. I don’t.

I would look into thin clients and Lenovo etc. tiny PC for office on eBay. I run old low power low noise rackmount Supermicros which are nice but hard to find at low prices.

Yes orders of magnitude, but not too many of them. The real estate of a 300 mm wafer is limited, the structure shrink is saturating and you can’t get too many layers. You still need a packet switched network on the wafer even if the rest is mostly analog. Perhaps spintronics can limit the power requirements too.

Factor in power bills and heat and noise into your calculations.

There are diminishing returns in semiconductor photolitho. Moore scaling is long over, absolute real estate see WSI with Cerebras, DC costs and power envelope are all sending a clear message. Quantization is there, so you can go from digital multipliers to analog and go spiking networks, but transformers and Co have little power there.

Also, the kind of economy that can carry Gen AI as business model is not a given, long term.

Neat. At 750 Wp I wonder what the EROEI of the installation is.

Potentially double use for space launches.

360 km or 194 nautic miles is the inter-port distance.

From the mouth of the beast https://developer.hashicorp.com/nomad/docs/nomad-vs-kubernetes What I have read online is that Nomad can have issues at large scale. No personal experience.

In any case since now OpenShift and Nomad are both under IBM’s umbrella there is space for an enterprise Kubernetes distribution, if someone is brave enough.

Nomad isn’t a real alternative to Kubernetes/OpenShift.

Question is also whether antichinese sanctions are contributing.

Full human equivalent across all capabilities. Gen AI fails pretty much everywhere.

There is no real AI out there, yet.

Good luck with that.