>How much energy does a car factory use? What about a chip fab? A hospital? A battery manufacturer?
tl;dr: rough napkin math and basic googling, 20 MW for a car factory, 100 MW for a chip fab, 2 MW for a hospital, and 0.2 MW for an apartment the same size as the data center. Water usage is equivalent to 14 car factories, 52 hospitals, or 18,000 people at home.
Good questions! You made me curious. In the process of looking, I came across this[0] reddit post with a chart comparing various industrial uses. For car assembly specifically, Honda has some nice global data here[1], and their total energy consumption is about 990 MW, split roughly equally between direct and indirect consumption. That's split across something like 60 factories! If I just look at direct energy consumption, that's less than 20 MW per factory.
This[2] has a chip fab at 100 MW. This[3] has an average hospital at 1.7 MW, with the largest use just being heating. I didn't try to look up battery manufacturers. This[4] has a typical apartment at 0.77 W/square foot, which would mean just 0.2 MW for an apartment the same size as the data center in the article.
I think a more interesting distinction is water usage. The linked article says that the Fireball LLC datacenter used 548 megagallons of water in 2025. Assuming that aligns with the "water consumption" rather than "water withdrawal" numbers from Honda, that's over 20% as much as the entire global operations of Honda equivalent to 14 factories (adjust by about a factor of two if we should be comparing to water consumption). Alternatively, it's equivalent to 52 hospitals[5] or 18,000 people in their homes[6].
Thank you. This seems a good start to a desperately needed credible "externalities report" to ground political and economic discussion.
The flood of anecdotal thumping seems to distract and undermine deliberation. The sad lack of peer reviewed data and impact analysis hinders thoughtful policymaking.
The Fireball one is concerning. 1.5 million gallons a day?
"More improper redaction shows that the data center using the most water annually is Fireball Group LLC, the Google Data Center in Papillion. Fireball LLC reports 547.88 megagallons for their 2025 Annual Water Consumption."
You had me curious about what I was tested for, also in the northeastern USA in the 2010s. It looks like they included Hep C when my PCP ordered STD tests in 2015, along with most of the tests the other commenter mentioned. I'm glad they caught yours in time to treat it well!
I suspect this is a real problem for charities, though. If those bots are using stolen credit cards, the "donations" are going to cost the charities money after they pay extra fees to the credit card processors. Nonprofits are sometimes used to test stolen credit cards before making more profitable fraudulent transactions, so there's a real risk of it costing them money if they get rid of the captcha but don't replace it with something sufficiently high quality, even after accounting for the occasional lost donation.
Merchants often pay a chargeback fee on top of refunding the main charge. Additionally, merchants with lots of fraud or other chargeback issues are likely to be dropped by payment processors or see their general fees with payment processors get more expensive.
Also, quadratics are just much easier to work with in a lot of ways than higher powers. Like you said, even powers have the advantage over odd powers of not needing any sort of absolute value, but quartic equations of any kind are much harder to work with than quadratics. A local optimum on a quartic isn't necessarily a global optimum, you lose the solvability advantages of having linear derivatives, et cetera.
I feel like there's a bit of a jump from "tech-savvy" to de-soldering things on an expensive piece of home electronics. As it stands now, though, I agree that turning off the smart TV features seems to be the way to go for most people.
Troy Hunt has been running Have I Been Pwned for years. He even uses the k-anonymity model to allow you to search if a password has been pwned without giving him the password if you don't trust him.
I get your general point, but he's been a leader in this space and walking the walk for a decade. I'm not even into security stuff or anything particularly related to this, and I still recognized his name in the OP domain.
More importantly, since HIBP sells monitoring services to 1Password, if they were maliciously collecting this data they would be immediately sued to oblivion.
I enjoyed reading the article, but I'm pretty thrown by the benchmarks and conclusion. All of the times are reported to a single digit of precision, but then the summary is claiming that one function shows an improvement while the other two are described as negligible. When all the numbers presented are "~5ms" or "~6ms", it doesn't leave me confident that small changes to the benchmarking might have substantially changed that conclusion.
Yeah. When your timing results are a single digit multiple of your timing precision, that is a good indication you either need a longer test, or a more precise clock.
At a 5ms baseline with millisecond precision, the smallest improvement you can measure is 20%. And you cannot distinguish a 20% speedup with a 20% slowdown that happened to get luck with clock ticks.
For what it is worth, I ran the provided test code on my machine with a 100x increase in iterations and got the following:
Which is not exactly encouraging (gcc 13.3.0, -ffast-math -march=native. I did not use the -fomit-this-entire-function flag, which my compiler does not understand).
I had to drop down to O0 to see branchless be faster in any case:
In general, modern compilers will often unroll or inline functions without people even noticing. This often helps with cache level state localization and parallelism.
Most code should focus on readability, then profile for busy areas under use, and finally refactor the busy areas though hand optimization or register hints as required.
If one creates something that looks suspect (inline Assembly macro), a peer or llvm build will come along and ruin it later for sure. Have a great day =3
Non-overlapping sub-problems may be safely parallelized, and executed out-of-order.
In some architectures, both of the branch code motions are executed in parallel, and one is simply tossed after dependent operations finish. We can't be sure exactly how branch predictors and pre-fetch is implemented as it falls under manufacturer NDA. =3
Can you link an example of a window unit ERV? I tried searching briefly, and came across some folks hacking together units to make them work with windows or adding their own ducting, but nothing analogous to a simple window air conditioning unit. As a renter of an apartment in a very much not modern home, I don't really see anything that seems like it would work.
A couple examples I see on Google. I'm not advocating for any of these, because I have no idea if they are any good, but I see no technical reason an ERV couldn't work as a window unit. Maybe it's an underserved market and someone should make a business out of that.
tl;dr: rough napkin math and basic googling, 20 MW for a car factory, 100 MW for a chip fab, 2 MW for a hospital, and 0.2 MW for an apartment the same size as the data center. Water usage is equivalent to 14 car factories, 52 hospitals, or 18,000 people at home.
Good questions! You made me curious. In the process of looking, I came across this[0] reddit post with a chart comparing various industrial uses. For car assembly specifically, Honda has some nice global data here[1], and their total energy consumption is about 990 MW, split roughly equally between direct and indirect consumption. That's split across something like 60 factories! If I just look at direct energy consumption, that's less than 20 MW per factory.
This[2] has a chip fab at 100 MW. This[3] has an average hospital at 1.7 MW, with the largest use just being heating. I didn't try to look up battery manufacturers. This[4] has a typical apartment at 0.77 W/square foot, which would mean just 0.2 MW for an apartment the same size as the data center in the article.
I think a more interesting distinction is water usage. The linked article says that the Fireball LLC datacenter used 548 megagallons of water in 2025. Assuming that aligns with the "water consumption" rather than "water withdrawal" numbers from Honda, that's over 20% as much as the entire global operations of Honda equivalent to 14 factories (adjust by about a factor of two if we should be comparing to water consumption). Alternatively, it's equivalent to 52 hospitals[5] or 18,000 people in their homes[6].
[0]https://www.reddit.com/r/aiwars/comments/1tes0rr/ai_data_cen...
[1]https://global.honda/en/environment/initiatives/data/
[2]https://www.mckinsey.com/~/media/mckinsey/dotcom/client_serv...
[3]https://www.eia.gov/consumption/commercial/pba/health-care.p...
[4]https://electricityplans.com/how-much-electricity-does-an-ap...
[5]https://pmc.ncbi.nlm.nih.gov/articles/PMC5058574/ for water usage per square meter for hospitals
[6]https://pubs.usgs.gov/of/2017/1131/ofr20171131.pdf for water per person per year