> He said YouTube Premium - its service letting users pay to remove ads between videos, or songs on its music service - had helped boost paid subscriptions across Google consumer services to more than 325 million in 2025 overall.
Out of the 60bn they made only 325 ml from paid subscribers. The title made it like it was an important figure. There's also no YOY numbers or profit so it's difficult to draw a conclusion.
And still no information about how much of the subscription revenue is shared with the content creators. Feels like YouTube is dumping all the money to shorts and other creative accounting, and giving nothing to the creators.
Recently a significant percentage of the folks I follow stopped making videos.
This is a great solution for a very specific type of team but I think most companies with consistent GPU workloads will still just rent dedicated servers and call it a day.
Other benefits: easy access to reliable infrastructure and latest hardware which you can swap as you please. There are cases where it makes sense to navigate away from the big players (like dropbox going from aws to on-prem), but again you make this move when you want to optimize costs and are not worried about the trade-offs.
I agree, and cloud compute is poised to become even more commoditized in the coming years (gazillion new data centers + AI plateauing + efficiency gains, the writing is on the wall). There’s no way this makes sense for most companies.
The advantage of renting vs. owning is that you can always get the latest gen, and that brings you newer capabilities (i.e. fp8, fp4, etc) and cheaper prices for current_gen-1. But betting on something plateauing when all the signs point towards the exact opposite is not one of the bets i'd make.
Well, the capabilities have already plateaued as far as I can tell :-/
Over the next few yeas we can probably wring out some performance improvements, maybe some efficiency improvements.
A lot of the current AI users right now are businesses trying to on-sell AI (code reviewers/code generators, recipe apps, assistant apps, etc), and there's way too many of them in the supply/demand ratio, so you can expect maybe 90% of these companies to disappear in the next few years, taking the demand for capacity with them.
I've began using Opus and I felt it was a class above all the rest. Used cursor and teste different models, but opus somehow was always much much better. Bought the max for 100$, totally worth it.
There is a level editor with the ability to show the optimal result for a custom level. In theory, one could recreate any official level and reveal the best solution that way. However, I haven't tried this to verify any intentional roadblocks by the developer.
on disk, so basically I'm trying to save the image of a solution and reuse it if the same quiz is required. So instead of recomputing the result just return the same image.
I think Cloudflare r2 has a generous free tier. You can also technically store images in redis I think. anyways, thank you for making this, really cool!
curious question from a non-programmer - are you checking against the exact same image (i.e. hashed), or is there an easy way of trying to match an image to a very similar one you've seen before?
Not OP, so I don't know what their website does, but there is a technique called "locality-sensitive hashing" that gives the same hash for similar items instead of exactly the same ones
Thanks! Emscripten is the tool that makes it relatively easy, for stuff with source code available. So I'd say pick something and try to build it with Emscripten.
For stuff without source code you'll have to use emulation. DOSBox and MAME have web ports already. I haven't looked into them much. But the Internet Archive has a big collection of games that run in them already. So check that to see if the game you want is already available.
I'm not sure how to objectively measure this but if you have an idea I'm happy to give it a try and let you know!
For the record I live in Taiwan and apparently we have some of the best internet access in the world. And, I get 5g basically everywhere in the country, even when I'm deep in the mountains.
But subjectively I can say the latency is usually unnoticeable at like 30mpbs, though I'm sure a pro fps gamer would notice. The latency from the Bluetooth controller is far worse. What is a little annoying is the I think compression artifacts? Whatever it is that causes blockiness when I move my camera around, that isn't really visible on static objects.
reply