Because Nvidia is personally investing in a huge amount of those customers, enabling the hyperscaling. If half of the circle breaks, the other half does too.
And those customers still have the demand to keep wanting more.
Contrary to the AI-bubble-doomsayers - most hyperscalers and neoclouds seem to be at capacity for current demand and even have already booked future demand.
That's... really wild. I'm amazed at how the characterization is basically a lie. I go to Planet Fitness where there are 10 TVs with different channels in a row and sometimes all the networks have the exact same story in the exact same light within minutes with each other. It's almost creepy how lockstep they are.
It's state dependent, but many states have no threshold. If you visit that state to attend a conference and you're on the clock or you do some work in an airport terminal during a layover, you need to files taxes in that state. Nobody has likely ever been pursued for this type of short term event, but they could.
> If you visit that state to attend a conference and you're on the clock or you do some work in an airport terminal during a layover, you need to files taxes in that state.
Please provide an example of a state with this policy - I've worked for giant corp that track city-level travel for taxation purposes and have never seen a situation where an employee had to file an out of state return for a conference or a one day trip.
In New York, for example, if you are there for less than two weeks your employer doesn't have to withhold taxes but the you must file a NY return and pay state taxes.
Your employer is keeping track of where you are working to cover their obligations. You are still ultimately responsible for your obligations. Any tax software will help you with this. TurboTax, for example, will ask if you made money in any other states last year.
Your new york link is terribly inaccurate (no surprise as it's trying to sell compliance services). The requirement does not apply to "some work in an airport terminal during a layover" - it applies to income from New York sources - per the state of New York's own website:
If you're working on a CRUD/web app, absolutely. There's so much data for web apps out there that LLMs now produce better web app code than most of my coworkers from 2022 or so, since it follows consistent style rules. Web app code all looks the same, anyway.
To be fair, overpopulation was never an actual concern. Since the 1800s, food production growth rates have outpaced population growth rates. Underpopulation _is_ a concern when infrastructure and social structures may not be able contract quickly enough to sustain our current quality of life.
As long as the population depends on non-renewable resources, technically any population size is overpopulation. With the technology we currently use our population levels can't be sustained for very long.
You can see all the bad things about modern cultivation but lack of resistance to climate change isn’t one.
Yes, for many swaths and types of traditional agriculture and fishing it will be doom, but this conversation isn’t about local ecosystem but global food production vs population growth.
Food production was never a problem, supply and management is.
Do you know why and how food production has been able to match population growth and how long much more it can scale? We are using almost all of earth's farmable land right now.
Not through farming a larger area of land, that's for sure. It's been mostly up to improvements in crops, nitrogen fixing, the invention of the reefer car, and other things.
Additionally, I'd like to remind you that we have a lot of wiggle room: if we stop growing alfalfa and stop doing stupid things like biofuel, that will free up an enormous amount of land.
While nearly half of all habitable land is used to some extent as agriculture, the total amount of land in active cultivation has actually been declining since the early 2000's and we currently abandon more farmland every year than we clear forest or convert to agriculture. Between 1992 and 2020, ~ 7% of all cropland was effectively 'abandoned' and much of the production capacity in the west is actually intentionally kept from being fully planted in order to control against overproduction of cereal crops which in the past has been justified by arguing that this control is necessary to prevent market crashes when distribution networks cannot absorb the full capacity of a harvest.
Globally, there are about a billion acres, 400ish million hectares, ~ about half the size of Australia, which sits unused according to FAO satellite and other observables datasets.
About half of the abandonment not due to cultivability at the simplest level, but rather that of transportation of the crops to market or of access to workforce (labor or equipment) to the land. Smaller, more distant crops without significant mechanization simply cannot compete at the prices possible through mechanization. With a lack of access to global niche markets and the higher prices for, say, premium organics, many successive generations of farming families have simply moved to town and left the land sitting, unused.
The other ~ half of abandoned land is because of degraded production capacity through mismanagement; either through over-application of petrochemical fertilizers instead of crop rotation, or through mechanized tilling without careful erosion management. Many farms have neither the time nor the resources (or in the case of large industrial operations the long-term financial foresight on the part of management) to engage in multi-year regenerative practices necessary to reestablish the soil mycostructure and probiotic ecosystems necessary to recover productivity.
However, if demand were to rise sufficiently and if production instead of distribution was the driving force behind food scarcity, we do indeed have arable land reserves which do not require further destruction of uncultivated wilderness to meet a growing demand.
There is actually an inverse relationship to land productivity at scale; the larger farms do extract more gross tonnage of product at lower costs of per-unit equipment and labor, but actually those operations significantly underproduce volumes of harvestable food if measured in 'per unit of land' instead of labor hours. Fields with multi-level, interconnected plant communities significantly overproduce in terms or total volume. This has been demonstrated in nearly every place on earth using very simple techniques that have varied names and different approaches to marketing, but are all very much the same thing: call it Fukuoka's "One Straw Revolution", Bill Mollison's "Permaculture", Historic Food Forests in Africa and the Decca in India, or in the Middle East "Oasis". In the Americas it still is "Three sisters" gardening. These techniques simply are not mechanized and delivering mass volumes of single commodity product into a global distribution network and that's where the difficulty and disconnect between global populations and available nutrition originates.
If there was a technological and market incentives to connect abundant and diverse harvests in rural communities with Bodegas, restaurants, and grocery store chains in the big cities without ten middle-men adding cost for the consumer without improving the product in the least, then half of the land abandonment issue could unwind and small farms could thrive, not by delivering bulk raw indistinguishable commodity to food manufacturing, but by delivering farm fresh produce to kitchens and fruit stalls.
If that were both low-cost for the consumer, and high earning for the farmer while having quality control and assured provenance then the conversation could center around making better use of even less total cropland and restoring wilderness to a near-native sustainable state.
But as it stands now, we're not short on crop land to human population ratio so much as short on effective distribution and incentives for responsible production.
/todays-soapbox
... this is kind of my area of peculiar skill and work. I've tried to steer multinational, multimillion Euro/dollar agriculture operations that produce significant volumes of global "organically produced" harvests in the direction of cooperating with rural and indigenous farming communities in conserving soil health to the benefit of all. That's been a mixed bag, so the forward momentum is with the local rice production in rural Yamagata. Same objectives, more cooperation and long term planning.
I think these are very fun and I would love to see more content you can visit with curl. I'm setting up a home chores + calendar + weather station that runs off my homeserver. I was contemplating making a plaintext version so very stupid devices could still read and display interesting data.
> Can this be gamed? Teaching the AI to write less like stock Claude and stock ChatGPT? Yes, absolutely. And honestly that would be great. Because half of the irritation is that all these AI-written article sound the same!
The biggest problem with plugging a generator into an outlet is you need a double ended cord. If you use everything correctly they work fine, but there are too many ways things can go wrong. The most obvious one is you plug everything in, start the generator, and the trip over the cord on you way back inside, thus sending the live male end flying toward you. Once in a while a kid will see these dual male plugs and plug them in - lots of ways things can go wrong here.
There is also the problem of backfeeding, but this is more theoretical - even if you know to limit your power use, your neighbors won't and so your generator will trip its own breaker. Linemen are taught to either treat power lines like they are live, or to short them to ground - it can happen they don't do this and you kill a linemen, but still not as big a worry as it is made out.
They make generator "inlets" that sidestep all these problems when used correctly. Most common I've seen is a block so you can't have the inlet and the main breaker both on at once.
> There is also the problem of backfeeding, but this is more theoretical
it is not theoretical from the POV of power company employees working on their lines. the power goes out, they go to fix the equipment or wiring, and if the inverters connected to the grid have not shutdown, there could be 10s of kW to even MWs being backfed into the system they are working on.
that's why in the US at least, grid-tied inverters are required to have protection against this. if they do not see power from the grid, they shutdown (and rather quickly).
Hang on. You didn't address the reason they gave in the half sentence your quote skipped. How are those backfeeders going to keep going when there's way way more load than they can possibly supply?
> if they do not see power from the grid, they shutdown
I realize this is a complete noob question but I'm still curious: with enough "backfeeders" could it not in theory be possible that the backeeders see each other's output and keep operating?
Say, a few houses at the end of the line, each with very little electricity consumption (no-one's at home), but with panels and inverters going full tilt, something like that?
Yes, this is a concern with some older inverter designs. They use phase-locked loops which are meant to lock to the grid frequency but if you and your neighbours on the feeder are generating and consuming a roughly equal amount of energy and the feeder loses its connection to the wider grid, the loops can lock to each other and form a stable island. How long this would last in practice is not really known, it would probably be rare for it to happen at all and even rarer to stay stable for more than a few minutes but it has been demonstrated so it could have become an issue. Newer inverter have specific anti-islanding protection to detect situations like this and disconnect if it happens.
(the specific case of very little consumption but panels and inverters going full tilt doesn't work as this would lead to the frequency rapidly increasing, exceeding inverter set points and the local grid collapsing with all the inverter tripping off. The danger scenario is closely matched load and generation, like a real grid)
Thanks, anti-islanding protection must be a fascinating topic. You could look for voltage, frequency, phase anomalies but you also have to balance between being safe and having too many false positives. I'm not sure I'd want to be a line worker knowing there may be many inverters, from various manufacturers and with varying levels of quality, in every direction.
Perhaps a more robust solution would be for the power company to install remote-switchable circuit breakers at various places of the grid. Use some sort of wireless IoT protocol to listen for "on/off" signals and to report back "I'm now on / I'm now off!". Then the power company can disconnect parts of the network by themselves, without trusting that every inverter, generator and battery will do the right thing.
Apparently the anti-islanding features in the inverters prevent this kind of situation. They need several minutes of clean power to start back up, and do other checks.
Again, all linemen are taught to short the wires or they are alive. All grids have regular places where two different lines meet and there is a switch at those points where in the worst case is they can cut the down line close a switch that normally is open and they can back feed a line from a different one and get power to people who are on each side of the break and then they can come back later to fix the break. These switches are miles apart, which means that you cannot be sure while you're working on the line someone else didn't go to that switch and change it.
Also, people have been back peeking the grid from their home generators for decades by now. Solar and inverters are a different variation of that attack, but it is nothing new. Any utility will have procedures because they can't guarantee that's not happening and we learned years ago.
I'm not saying it never happens, only that correct procedure already makes it not an issue
It's not so much being unplugged as it is sensing whether the mains power is down. If it is, they stop backfeeding immediately, just like hardwired solar inverters do.
If this weren't done, there would be risk of electricution to anyone working on wiring in the house or on the utility distribution network.
Yes, the "balcony" style setups, sync to the power outlet of the house, which could be the grid, or it could augment a generator when the grid is down.
Furthermore, the danger of backfeeding is if you do it in a power outage and you start to power your neighbors' houses. It'll overload your generator and create a hazardous situation for anyone working on the wires. (This is why you're supposed to install either a generator switch or interlock. I have one that only allows me to energize when my main breaker is turned off.)
But how do they do it technically? Some countries require anti-islanding devices for grid connected solar, these are around $1000. That alone is much more expensive than the $300 plug in panels.
The inverter can sniff the grid power at frequent intervals to determine its voltage and phase.
It outputs at the same phase and frequency as the grid, and with a slightly higher voltage (think of a ball rolling down a slope from a higher voltage to lower voltage).
If the grid is down then it stops the inverter for a minimum safety period (some minutes iirc)
Generators are typically grid-forming. They can (and will) be out of phase with the grid power, and will backfeed and create hazardous conditions upstream if the grid is out (and your main breaker is still on.) To use a generator safely you need a transfer switch (or similar) to make sure only one source of power can be active at a time.
Solar inverters (esp for this use, but also in general) are grid following; they'll use the existing frequency and support it, and if the grid goes off they'll go off too and do no power-off backfeeding.
The inverters are thus broadly safer to be plugged in to systems that are otherwise unprepared.
Generators could be made grid-following (many are now inverter-based), but for most users that would be silly, since the supplemental power would be more expensive (and louder!) than from the grid.
In addition to the electronics that others mentioned, they also have limited amperage. That way even if you are pulling from your mains power and the solar at the same time, your circuit hardware shouldn't melt or catch fire.
Generators are synchronous machines which means they can produce a useful reference frequency without any external source as long as they are spinning at a reasonably correct RPM. This RPM is easy to enforce with mechanical means. This independent operation is also what makes them so dangerous. There is no control signal from the grid back to the generator.
It can be quite safe to plug a generator into an outlet if you take precautions to properly island the system. I had an interlock on my breaker panel that would let me turn on a 50A backfeed breaker if my main breaker was switched off.
The back feeding problem is actually identical for plug-in and regular inverters. When the grid goes down, they keep generating power, creating a disconnected 'island'. All inverters are designed to be incredibly unstable in this mode, quickly desynchronizing from each other.
tl;dr Plug-in solar inverters must power off within a second if they stop detecting the grid's sine wave.
Generators have no such protection and will happily backfeed through the transformer outside your house to energize power lines. Line workers are careful, but there have been deaths. After hurricanes seems the most common based on a quick search.
So it's not a stupid question, indeed you should never plug a generator into an outlet. The double-ended "suicide cord" which allows you to do that has that name for a reason. (It can be fine to plug a generator into an inlet circuit, if the electrical panel has a proper interlock that physically forces disconnect from the utility before connecting the generator inlet.)
Yep, if you're gonna redneck the generator after a hurricane, you'd wanna disconnect your house from the grid. One trick in more rural areas is to have your power meter hooked up to one panel, and then your house panel be a "sub panel" to this main panel. Now you can shutdown your main panel and still redneck your house.
The better way (and legal way) to do this is to have a generator plug plug installed, this way when you switch over to generator your property is disconnected from the grid.
reply