They're laying out their plans for up to the end of the decade. You think this is just a fraction of their NA plans, that they're hiding things from their giant "Look, We're Doing EVs!" press release?

Creating the training dataset is the *last* step. I have dozens of TB of raw data which I use to create training datasets that are only a few GB in size. Of which I'll have a large number sitting around at any point in time.

Take a translation task. I start with several hundred gigs of raw data. This inflates to a couple terabytes after I preprocess it into indexed matching pair datasets (for example, if you have an article that's published in N different languages, it becomes (N * N-1) language pairs - so, say, UN, World Bank, EU, etc multilingual document sets greatly inflate). I may have a couple different versions of this preprocessed data sitting around at any point in time. But once I have my indexed matching pair datasets, I'll weighted-sample only a relatively small subset of it - stressing higher-quality data over lower quality and trying to ensure a desired mix of languages.

But what I do is nothing compared to what these companies do. They're working with common crawl. It grows at a rate of 200-300 TB per month. But the vast majority of that isn't going to go into their dataset. It's going to be markup. Inapplicable file types. Duplicates. Junk. On and on. You have to whittle it down to the things that are actually relevant. And in your various processing stages you'll have significant duplication. Indeed, even the raw training files... I don't know about them, but I'm used to working with jsons, and that adds overhead on its own. Then during training there's various duplications created for the various processing stages - tokenization, patching with flash attention, and whatnot.

You also use a lot of disk space for your models. It's not just every version of the foundation you train (and your backups thereof) - and remember that enterprise models are hundreds of billions to trillions of FP16 parameters in their raw states - but especially the finetune. You can make a finetune in like a day or so; these can really add up.

Certainly disk space isn't as big of a cost as your GPUs and power. But it is a meaningful cost. As a hobbyist I use a RAID of 6 20TB drives and one of 2 4TB SSDs. But that's peanuts compared to what people working with common crawl and having hundreds of employees each working on their own training projects will be eating up in an enterprise environment.

Taiwan reinforces its silicon shield.

This is all to produce a peak of 240k EVs per year. Production "starts" in 2028. It takes years for a factory to hit full production. Let's be generous and say 2030.

Honda sold 1,3 million vehicles in the US alone last year - let alone all of North America, including both Canada and Mexico. If all those EVs were just for the US it'd be 18% of their sales, but for all of North America, significantly less.

In short, Honda thinks that in 2030 only maybe 1/7th to 1/8th of its North American sales will be EVs. This is a very pessimistic game plan.

And you are the gauge the whole world as to compare to.

It's pretty easy to offer consumers a great deal when you're willing to dump wheelbarrows of cash into a furnace to do so.

As this would amout to cutting people off from their ability to make phone calls, and the people aren't at fault in this case, this will not happen. And no, leaving emergency numbers available is not sufficient, because people might have to call their relatives, their lawyers, their physicians or other. non-emergency-numbered contacts.

Ford made the Ford Ranger EV 1998 to 2002, then the Ford Focus Electric from 2011 to 2018 before switching to the Mach-E. They are not "new at it". They're just bad at it.

To be fair, I have a lot more hope for Ford than GM, as Farley seems to actually understand the critical importance of turning things around and the limited timeframes to do so, unlike GM, which still seems to only care about press.

Copper is not "the last mile". It's the last five meters. If that. When people talk about "the grid", they're not talking about the wiring in your walls. Which you don't have to redo anyway for adding an EV. Nobody has to touch, say, your kitchen wiring to add an EV charger.

"The grid" is the wiring leading up to your house. Those conductors are alumium, not copper. Occasionally the SER/SEU cable will occasionally be copper, but even that's generally alumium these days. And that's only to the service connection point (not even to the transformer - to the point of handoff between grid-owned and the homeowner-owned, generally right next to the house), e.g. after the service drop line with overhead service that descends down to the building. The "last mile" is absolutely not copper. Approximately zero percent of modern grid-owned wiring is copper, and even the short customer-owned connection from the drop line into the house is usually alumium.

Grids are not copper. Period. This isn't the year 1890 here.

And no, grid operators don't make money selling power. They make money providing the grid through which power is sold.

I have never seen a single utility that charges a flat grid access fee to residential consumers, anywhere on Earth.

Distinction can be hard to grasp for someone utterly ignorant on the subject

Says a guy who thinks that there's a mile of copper leading up to your house.

Not in this case. If someone tells me, that he mistrusts Cisco products and would rather not use them, that's fine. If somone tells me that "even he knows better than to use Cisco", I have to assume that his qualification is that of an armchair expert.

If someone's understanding of network security consists of "Don't use X", I know I will never hire him for any consulting work.

California does not have a history of major budget deficits. It has a major deficit this year, but has had recent surpluses. Over 25 years it is close to balanced most years and about as likely to be in surplus as deficit. This year is an anomaly - not a pattern. Look at the data.

Grid maintenance is done regularly. There is a very close link between the grid and power generating companies that sell power -- much closer than the link between any commercial user of highways for example. Decades of deferred maintenance does not happen on power grids since that would prevent the generating companies from selling power. They invest, and see to it that there is investment.

Maybe you should read the study. It obliterates your fake statistics.

Electricity grids require an maintenance investment of about 2.5% a year -- you effectively replace the grid every 40 years. If the spending is tripled during the build-out, then 1/3 of that is the grid maintenance cost that would be present anyway. After the EV build-out finishes the cost would be 2.5% of a ~50% larger grid.

The headline was the most extravagant sounding aspect of the report "cost up to 20 billion dollars!" instead of highlighting the conclusion "electricity costs will remain stable".

There will be no EV apocalypse.

BTW from 1960 to 1980 electricity demand increased 5% every year - a similar period of rapid grid expansion. And the major driver for that were air conditioners which were rare before 1960, but all but universal by 1980 -- everyone who wanted them had them. And electricity demand when flat in 1980 and stayed flat for decades after. We have seen this scenario before.

The grid is not made of copper. You thought it was? Copper is for home wiring, if that. Up to that point, it's alumium, bundled with steel on major lines for tensile strength. Does it look like copper to you?

As for the article: grid operators don't build out grids on a lark. They do it to sell power, because they make money selling power. If people want to buy more power because they want to charge an EV, then that's more money available for them. EVs are a boon to grid operators. They're almost an ideal load. Most charging done at night, steady loads, readily shiftable and curtailable with incentives, etc. Daytime / fast charging isn't, but that's a minority. And except in areas with a lot of hydro, most regions already have the ample nighttime generation capacity; it's just sitting idle, power potential unsold. In short, EVs can greatly improve their profitability. Which translates to any combiation of three things:

1) More profits
2) A better, more reliable grid
3) Lower rates

    * ... depending on the regulations and how competitive of an environment it is.

As for the above article: the study isn't wrong, it's just - beyond the above (huge) problem - it is based on stupid assumptions. Including that there's zero incentives made for people to load shift when their vehicles charge, zero battery buffering to shift loads, and zero change in the distribution of generation resources over the proposed timeframe. All three of these are dumb assumptions.

Also, presenting raw numbers always leads to misleading answers. Let me rephrase their numbers: the cost is $7 to $26 per person per year. The cost of 1 to 5 gallons of gas per year at California prices..