I agree in principle with the math. But I believe that in reality if revenues don't show up quickly, then lenders will just restructure the debt and defer the payback period. Similar to SF commercial real-estate; many buildings should've come due during the depressed covid market, but lenders (banks) were willing to delay payment until the market picked up again.
The scale of these investments put the lenders at substantial risk, so the lenders will do anything to make it work. If the current lenders will be damaged by extended payback periods, they can simply sell the debt to someone else who won't be.
Dario (the founder) has a phd in biophysics, so I assume that’s why they mention biological weapons so much - it’s probably one of the things he fears the most?
@deno team, how do secrets work for things like connecting to DBs over a tcp connection? The header find+replace won't work there, I assume. Is the plan to add some sort of vault capability?
WASM for frontend, at least, has been held back by fundamental tools like bundle splitting, hot-reload, debugger symbols, asset integration, etc. We spent a lot of 2025 working on improving this. Vite and friends are really good!
I've been working on a big Dioxus project recently and am pretty happy with where WASM is now. The AI tools make working with Rust code much faster. I'm hopeful people gravitate towards WASM frameworks more now that the tools are better.
The browser UI will likely be more of a cool demonstration of the project instead of the end goal. We want blitz to exist to help make it easier to build stuff like lightpanda. There's a whole world of interesting browser forks that could exist but don't, and being able to easily remix the browser opens the door to new stuff like AI automation, hybrid native gui frameworks, better accessibility tools, etc.
That's why Rust was introduced into Firefox piece by piece. The goal wasn't to rewrite firefox in Rust - just to migrate the scary bits over to a memory safe lang. You can feel a lot of that in the servo codebase, weird pointer semantics as a result of needing to be API compatible with the C++ adapters.
If I were building a company around a new browser, I'd reach for the solid stuff that can be pulled in. Our whole blitz project is designed to be modular exactly for that use-case.
I mean you can opine about how Rust isn't suited for browser development, but as someone building a browser in Rust, I think it's just fine. If anything, Rust has been really shining in this project since Rust was designed to build a web browser.
Also I think it's a little ridiculous to build yet another new browser in a new language when so many amazing pieces are just sitting around ready for someone to use. Come contribute, we're already much further along :)
It seems like the economy is on a “K” shaped flywheel. How much worse can the economy get for the regular worker before the systems just pops? We’ve put so much speculation into an AI/tech salvation that seems premature, especially when you look at ROI vs depreciation timelines.
I’m not sure what timeline to place on that but there has to be a floor for how bad it can get for the regular man.
Shit is just expensive. Young people can’t buy houses, good jobs are drying up, and inflation isn’t stopping.
I think a hallmark of 2025 is a resounding lack of empathy and compassion from people. Maybe's it's smartphones, social media, or some sort of existential doomerism.
To reframe your scenario: imagine you went to a school and some of your classmates came from poor families and couldn't afford clothes, food, or a laptop etc. To help those students, the teacher used class funds to buy them new shoes and get them a nice laptop to get their work done. Do you still think it's unfair that you don't get new shoes, laptop, or cookies?
The solution to your original question is to understand why the teacher is giving girls 4 cookies and then just be happy that more people get a fair shot at life.
In high school, I ran a robotics team that did lots of STEM outreach. We went to community centers, after school programs, and worked with other similar orgs like "girls who code."
I think we played an important role in the community. In our mission we stated we wanted to help bring "equity to STEM education."
In 2025, according to the current admin's stance on "DEI," my robotics team would not be able to receive grants without risk of being sued. It's plainly obvious the line is not drawn at restraining "overly progressive policies" - it's just arbitrarily placed so the govt can pick and choose the winners based on allegiance.
It's a shame that folks with a strong moral fiber are now punished for wanting to help their communities.
I think this might be the “it” moment for AI/LLMs. I was hiking with a friend recently and we talked about this at length.
The arc-AGI results from O3 are apparently a result of chain of thought given enough time to explore a solution space. Reasoning might be simply a higher dimensional form of rubix cube solving. BFS, search, back-tracking, etc. It seems unlikely that humans think in “tokens” so why do LLMs?
By staying in latent space, the models are free to describe an “idea” in higher resolution than what language allows. English is coarse, granular. Latent space is a much finer representation of ideas and their interplay.
Latent space is also much cheaper to execute in. The model can think without the language encoding/decoding step. This lets it branch out hundreds of ideas and explore only the most useful ones in a fraction of time that reasoning “out-loud” would take.
The states also don’t need to be tied to language. Feed in a robot’s state, time series data, or any abstract data. Reason in category theory or linear algebra or complex analysis. Humans are hard wired for one set of math - an abstract latent space can represent anything.
I’m a bit disappointed OpenAI didn’t stumble on this first. I’ve been skeptical of LLMs since their big debut last year. LLMs seem like a great way of solving language, but reasoning is much more complex. Once you grok the math behind the current models, you immediately question why the encoding/decoding step is there. Diffusion models are incredible but it felt that LLMs lacked the same creativity. Encoding/decoding forces a token-based discretization and therefore a loss of complexity.
With the byte-latent paper it was quite clear we’d see this paper. This truly might be the “it” moment.
- Opus 4.7 xhigh: 5.2%
- Opus 4.8 xhigh: 13.4%
- Fable 5 xhigh: 29.3%
Seems like a huge jump.
[1] https://cognition.ai/blog/frontier-code