I don't want to argue here, but once models are trained, using them is quite cheap. Also in the last few months most improvements I've seen are in the smaller sized models, which might make using models even cheaper.
If the models are cheap to run, while good for customers, isn't that a problem for hyperscalers and all the investment poured into the AI data center build out?
No. If models become better, the natural expectation is for hardware to get more valuable, not less. (The hardware that was previously able to achieve X can now do 2X. Why would that make the hardware worth less?)
Likewise if models become commoditized, it's raising the value of the existing hardware (you'd expect the hardware owner to keep a bigger slice of the surplus when that's the scarce component).
For model improvements to be bad for the value of hardware, what you'd need is for demand for intelligence to be inelastic. But everything we've seen so far suggests that is not the case: the demand for intelligence (whether human or artificial) is highly elastic both in price and in quality.
its a false equivalence.... the dotcom boom wasnt just a fibre rollout, it was all the internet company startups that were overpriced and crashed. no doubt some of the hyperscalars will get busted out, but the models and LLMs are here to stay. just like dot com websites didnt disappear after the dot com crash, neither will AI transformer models
There's this weird tension where small models are getting a lot better, but the gap between the frontier and "what I can run on my gaming GPU" at home isn't necessarily getting smaller. That said, Qwen-3.8-27B-Q3 runs on my 9700 XT and performs well enough for tool calling in agents, which is a critical floor in local model performance imo.
Do you use qwen for any coding? Do you make a frontier model orchestrate it? Are you running with llama.cpp? Sorry for the barrage of questions, but I've been interesting in local AI for a while, yet I've struggled to set up anything usable.
My Qwen testing has been with Hermes trying to do mundane (to me) things that hosted models understandably refuse to do, even open-weight ones like Deepseek V4.1 Flash. Example: I tried to get it to find a torrent for $tvshow and add it to qbittorrent (connected to hermes via MCP already). It certainly tried and I think frontier models would have done better. My biggest issue was context space more than anything else. I had to drop down a quant level to increase it to 128k. On 64k it was constantly compacting.
That's a big issue in my team. We cannot really review tests manually, we already have too much just reviewing the code. And even then, sometimes the 'is this change necessary' question is forgotten or ignored (happened to me 20 minutes ago) because I an tired of fighting my junior+AI. I try to make AI refactor the tests every weeks ande that's it.
I use soverin since I moved from gandi 3 years ago. I didn't had any problems in this time, but I also didn't registered my domain with them just in case.
I don't follow the car industry but I've worked on a project circa 2018 for all European manufacturers to help provide parts 15 years after the car was sold (I provided the Hadoop clusters and some of the predictive algorithms to try to predict which car parts will fail and how many will be needed over the car Lifetime.) BOOST4.0 or something. I don't know where it ended up going.
How stiff can you make the resin with those 3d printers ? Probably not as much as carbon fiber, but how does it hold against aluminium or fiberglass? Also, can you do composite fibreglass/resin with 3d printers ?
Surprisingly complicated question. Most hobbyist grade machines are low powered and can only cure resin with pretty poor mechanical properties. Industrial resin printers can actually make incredibly strong parts.
The studio 1 can apparently print in Formlabs Tough 1500 which is a decent strength resin. It’s noticeably better than the average “ABS like”.
Resin prints are also isotopic (unlike composites and fdm prints) which has a lot of advantages for mechanical parts.
Asking if resin prints are strong is a bit like asking if metal parts are strong; it really depends on the metal or resin being used
I think you misread. The article state tesla Semi has had 2 fire in 2 years. The first one because the truck hist stuff and left the road in Nevada, the second (this one, the subject of the article) in California for unknown reasons. This is relevant because the Tesla semi fleet is small (86 vehicles 2 years ago, probably below 500 atm) and because battery fire are hard to stop, so we want to know what caused it. If EV fire last 10 time longer than diesel fire, but happen 10 time less often (we need more studies but it seems this is the initial consensus), this is not really an issue, but if they happen more often, we _need_ to know why and fix it.
Also i don't want trucks to catch fire everytime they hit a cow or a boar, the roads around where i used to live would be consistently closed :/
This vehicle was part of a manufacturer-owned highway fleet. That means it received proactive maintenance by trained personnel on a regular schedule, with full access to engineers for diagnosis and OEM replacement parts, and ran on good roads in the mild California climate. It's practically pampered compared to oilfield haulers, especially the ones that live long enough to be auctioned off. They're not comparable.
As an aside, some of the vehicles I've worked on serve in oilfields and I'd be horrified to discover 1% of them were catching fire annually.
For some other unscientific findings, that'd suggest 5 vehicle fires per year just from residents of my apartment complex. I've seen 1 in 8 years, and that was arson.
The fires/mile ratio seems like a good idea (skewed a bit by relative driving habits of professional vs typical drivers). It turns out those numbers are all pretty readily available. All miles from all sources (including semis) tally to 3.3T/yr. All car fires from all sources (including crashes, so in the context of semis spontaneously combusting these numbers will make the 1/100 rate look even better than it should be, though separate sources indicate that most car fires aren't from collisions, so that's perhaps a small effect) total to 210k/yr. You then find that, smoothing out all underlying subgroups and data, there's a fire every 15.5M miles.
Applied to residential apartments, there's an average miles per year of 14k or so (should be less here than average based on anecdata, high fuel prices discouraging driving, and everything being close even if you do drive and usually close enough to be walkable, but let's assume 14k/yr is a reasonable baseline), giving us a fire every 1100 vehicles, not every 100.
1/1k is more plausible to me (still "feels" high, but it doesn't trigger the baseline absurdity checks I have in the back of my mind). 1/100 suggests that every year in uni one of my classmates should've had their car burn down. 1/100 suggests that every year my hometown newspaper should report 30 cars burning down -- one every week or two (and given that the newspaper covers the entire county it should actually be 3/week). It's a pretty high rate. The typical person knows hundreds of people well enough that a car fire would often crop up in conversation eventually, and I don't think the typical person knows anyone whose car caught fire. Elaborating on that, I don't think the typical person doesn't know somebody whose car caught fire _ever_, much less in the last year.
In my personal experience pretty much every old farmer has had multiple vehicle fires. It's quite likely location dependent -- if you're in an area which clogs up air filters quickly, and or where dry straw getting in where it shouldn't is a likely experience, vehicle fires are very common.
Maybe. That hometown I referenced has 3k people in the city (not farmers) and 12k people in the county outside the city (mostly farmers in varying capacities). It's the only place I've lived where it's common for people to cut, mill, and dry their own lumber for economic reasons, or (ignoring parts of Alaska I've lived in) for hunting to be an economically rational thing to do. There are literally zero vehicle fires. A home or business per year or two burns down, but at the 1/100 vehicle fire rate we expect 3 fires per week, and there are zero in the last decade. It's a hot, dry climate made up of mostly farmers with extremely shoddy equipment.
Anecdotes aren't everything. If there are 1/1k fires annually then those have to come from somewhere, so your "location dependent" idea makes a lot of sense. I'm passionately curious now as to where those fires actually happen though. It's not just farmers with bad equipment in bad scenarios.
People can say it's 'years of neglect' and might even believe it, but to me it's clear train liberalisation directly caused 2 train incident in Spain this year. Yes trains are slightly cheaper (for now, look at GB to estimate how long the low prices will last), but I'm not so sure the redundancy you pay for in a public monopoly is worthless, especially in areas where an incident can cause this many deaths.
"Directly" is way too strong. Yes, it may have been a factor here that the train that remained standing still was from another operator, but there are many more important factors at play, the main one being apparent lack of safety mindset of the controller. In other European countries with even more operators, the safety record is not commensurably worse.
It is though. It's like the electricity market. We try to liberalize it, we see it doesn't work, so we add regulations and regulators (here the "controller"), and each time there is a failure, we blame regulation or regulators, or both. Sometime when it's too egregious we blame the particular company that was caught fucking up, because blaming the system that allows such fuck ups would cost to much: ideologues in power agreeing to the idea that maybe, market do not work everywhere.
USAID was a vehicle for US intelligence. That said, it also helped a lot where it operated. Especially helped logistics and local administration tasks for NGOs which could be extremely bad at those two tasks.
Also USAID existing prevented NGOs to have too much local political power and the corruption that goes with it.
My opinion is that overall, USAID was a net positive, although in some specific cases it wasn't.
USAID was absolutely a net positive and if nothing else its elimination has completely ceded almost all soft power to China and Russia, something that will have repercussions for decades to come.
> A rise in instability abroad also creates a national-security risk for the U.S., Wright adds. Yemen—once a major USAID recipient, with $768 million worth of aid flowing into the country in 2024—is also home to the Houthis, an Iran-backed militant group that controls much of the country’s north and has been designated a foreign terrorist organization by the U.S. As USAID funding disappears, support for the rebels is only likely to grow, he says. “Iran can continue to finance rebellions like the Houthis,” says Wright. “These second- and third-order effects might be even more consequential to the U.S. than the direct effects.”
I would be interested in reading some reporting about USAID's role in support of the CIA (or DoD intelligence?). Please recommend some that helped form your opinion.
I have, luckily, read a personal account of the logistical assistance that a USAID contractor had [0]. Doug Muir wrote about improving customs practices in Rwanda and training locals to take over for him. Improving, in this case, means standardizing inspections of trade goods to decrease holding time.
Also declassified files about mkultra and mknaomi mentioned USAID but I didn't read the specifics so it might not be that bad.
But yeah, I've never read the book you're talking about, but USAID usefulness wasn't so much in medicine and food (which are still important), but in organisational, logistics and administrative support to NGOs that really increases their effectiveness, and especially small NGOs like libraries without borders or OLSF( my cousin directed this one for a few years, that's how I know how useful USAID was)
Yes. And bear in mind than default create inflation naturally. When it's productive companies that default, the effect is increased. If PE can't roll their debt, i fear it will increase the dollar inflation really fast, which will have second order effect.
Also since idiots predicted hyperinflation 5-4 years ago and other idiots believed them, we're still before the 'bust' of precious metal prices (We're at the point where people invest in opening new gold/silver mines), which means it will be harded to edge against inflation than it used to be.
The Fed has been trying to hit its 2% inflation target (which is arguably unrealistic as the neutral rate but that's another topic) for half a decade and can't hit it. I suppose the question is: What do you consider hyperinflation objectively?
Due to structural demographics and other factors (exhausting institutional buyers of sovereign debt no longer required to buy and hold it at below market pricing, and sovereigns now having to go sell debt into a price sensitive debt market), we've gone from a capital surplus to a capital shortage, and this will likely continue long into the future due to demographics (working age population shrinks, wages get pushed up, and a whole lot of debt that needs to get refinanced at current and forward curve interest rates [real estate, private equity, sovereign debt, etc]).
If you can forecast the cost of money curve, you can forecast "who has been swimming naked when the tide goes out" based on their debt load and future cost of debt, and therefore when the economics turn against what the fundamentals can support.
I'd say over 30% monthly for at least 3 months or over 8% monthly over two years are good targets to call something hyperinflation. Different kind of hyperinflation have different causes and resolve very differently.
2% yoy inflation when assets grow 10% yoy mean reducing the purchasing power of work. Inflation target should be around asset growth if you choose to tax income rather than wealth.
reply