9cb14c1ec0

Or you live in Maine with no datacenters and see massive price increases anyway, because God forbid we actually had a reasonable energy policy in the state.

MAAS

Question for the experts: does the power crunch mean that AI hyperscalers will turn off previous generation GPU datacenters to free up power for their new Vera Rubin GPUs?

Nah, a friend's fiber bill just went from $50 to $100 per month. He is switching to Starlink.

The genie is out of the bottle, folks. You can find some pretty good vulnerabilities even with models like Deepseek V4 Flash.

If it actually was spent on ads, it seems to me OpenAI would have to be one of the single biggest ad buyers in the world. Almost certainly they are using a somewhat broad definition of sales and marketing to cram a bunch of expenses into it to make some other category look better.

Very urban perspective. If you are a contractor or farmer in rural America, there is no way to survive without a good sized truck.

I believe it. Makes me curious what your prompt was that got such a good result out of Opus.

Yes, we have successfully used Mistral OCR for digitizing handwritten forms. You always have low percentage that need human review and adjustment, but overall Mistral has been highly accurate (their price is amazing, too).

Google Play has been a huge drag on innovation and security in the mobile ecosystem. I'm actually looking forward to the time when AI kills the mobile app ecosystem so that every phone manufacturer can bundle their own "vibe-code-your-own-app" system with their devices, and the Google Play monopoly is broken.

Meanwhile Deepseek V4 Flash will happily hunt security vulns at almost 0 cost. We are ceding the bug hunting to the open weight models.

It turns out the demand for software is so great that even when it can be produced instantly at zero cost demand still far outstrips supply.

I spend about $100/month with Vultr. The uptime in the datacenters my VMs are hosted in is extremely good.

The can't afford it exception is disappearing soon, as it isn't true for any business. Total setup costs for STIR/SHAKEN are under $2000 these days. Providers that lack control over the network infrastructure (i.e. they don't have the ability to control the stir/shaken headers so by definition they can't spoof numbers) will likely continue to be a thing as changing it would force pretty much every small business in the VOIP industry out of business and allow only large companies to be VOIP service providers.

STIR/SHAKEN up to this point has only been a self-certification that a telecom company has the right to use a number. What the FCC is trying to do is set up a legal obligation for the STIR/SHAKEN header to match a KYC verified identity.

If the FCC implements this, I expect a lot litigation because of the burden and legal liability this would place on telecom and VOIP companies. There are other less burdensome approaches to preventing spam that the FCC has not tried.

Because the concept of numbers is so heavily baked into many systems. Momentum is a beast.

I do not believe for even a second that Telecoms Grade switching equipment is unable to do the same

Mikrotik is a young spring chick compared to the dinosaurs in telecom.

That's because they are bulk purchasing numbers from voip providers, cycling through probably hundreds per day.

The code I get out of Opus 4.8 is better than the code produced by most programmers I've met.

There are two ways to challenge FCC decisions. There is the upfront approach where a business whose operations are harmed by an FCC decision sues to block the decision. Then there is the approach where said business announces their non-compliance and dares the FCC to sue them. The FCC does not have criminal charging authority, so it has to rely on courts to enforce compliance. See the Federal Communications Commission v. AT&T case that just wrapped up at the Supreme Court.

I hear you, but with the hype surrounding Mythos the demand is going to be insane. I'm already hitting server errors in claude code.

I expect the FCC to adopt this rule, and I also expect it to be challenged in court, on the basis that there are many other approaches to fighting spam calls that the FCC has not tried, but are much less intrusive.

So we know what they are renting these GPUs for. I'm really curious about the input costs of their power generation. Is there actually enough margin in these deals for xAI to cover their depreciation cost?

Edit: from the footnotes: > Colossus actually runs largely on its own on-site gas turbines, which comes out even cheaper: at a simple-cycle heat rate of ~10,000 Btu/kWh and Henry Hub gas at ~$3.50/MMBtu, the fuel bill is only around $90mn a year.

OK, that's crazy. How can I get into renting GPUs to hyperscalers?

Anyone remember the old days when a new frontend framework came out every 3 months. That has pretty much stopped. No one cares anymore.

As per usual arrangement, the internet can't stand a nuanced opinion, but instead jumps straight to extreme conclusions. Nowhere did I say anything about wallowing in poverty.

We need universal childcare services

Birth rates have been falling worldwide, regardless of the level of government support. It's much more a matter of attitudes about having children.

The economic costs of having children at a replacement rate are simply too high

Nope. My wife and I have 4 children, on a lower-middle-class income in the US. Your lifestyle choices matter a lot. If you want to have children, you can find a way to afford them.

I once submitted a PR to Ladybird, but even in early AI days there were so many open PRs that mine got lost in the noise. I don't really blame the maintainers here. Once the open PRs get to a certain point, it becomes unmanageable.

Fortunately RAM is largely a commodity and more for sale only in China means more for sale elsewhere.

That doesn't mean it can't be used as a market signal. These 2 things can both be true at once.

Do other budget brands really print as well as Bambu out of the box with no tinkering?

Yes, everyone has caught up to Bambu these days. I have personal experience with Creality and Anycubic.