When I need a break from my desk on the weekend, I go on a 2 hour walk and just work by voice noting @instinct
Incredible feeling walking along this coastline while smashing things off my to-do list
It's massive. Bigger than incumbents imagine. Can confirm that most execs working in Windows, CoPilot and MacOS are really not using intelligence. Their main touch point is something as meaningless as Muse or Instinct.
Attack.
Marc Andreessen says the gap between what AI-native startups and incumbents think AI can do is the widest he’s ever seen:
"The AI-native early stage startups have just a completely different view on what they can do with AI than the typical incumbents in almost any industry, almost without exception."
"AI is just racing ahead in capability even week by week... There are going to be a lot of industries that are really up for grabs in a fundamentally more profound way than maybe has been the case in the past."
"The founders are gonna have a level of superpowers coming from the use of these tools and these technologies that they just simply haven't had in the past... Had I had Astra and Netscape in 1994, like, oh my God. Like, Ben, what would you and I have been able to do at Netscape?"
"Those entrepreneurs are gonna color way outside the lines. We already see this. They're not gonna think of themselves as like, 'I'm just a software company,' or, 'I'm just a hardware company,' or even, 'I'm just an AI company.'"
@pmarca@bhorowitz
In the end the labs will own the application layer outright. The hardware and the operating system stay out of their reach, and that is the only ground left to hold.
Assume any application a frontier model can already operate through computer use is on a clock. Blender, Premiere, Excel, Figma, all of them. Computer use exists because the model has no other way to get the work done... YET... so it looks at the screen and drives the app the way a person would. That is a bridge, and bridges get retired once the direct route is built.
Blender’s value is its capability: modeling, rigging, rendering, simulation. The interface is just how a human reaches that capability. Once the model can produce the geometry and the render directly from the request, the viewport, the menus, and the plugin architecture become dead weight. All the model needs is the output. And every hour the model spends operating an app through computer use is a lesson in what that app actually does, which is exactly the capability the labs will build into the next model.
This runs across the entire application layer. Image editors, audio workstations, IDEs, office suites, 3D and CAD. Each is a bundle of capabilities wrapped in a UI built for human hands. The model absorbs the bundle and drops the wrapper. Third-party applications get eaten.
What the labs cannot absorb is the layer underneath. They do not own the silicon and they do not own the operating system. The model runs on hardware someone else manufactures, inside an OS someone else controls, and that OS decides what the model can see, touch, and do on the machine. Everything above that line is software a model can reproduce. Everything at and below it is the machine itself.
Intelligent PCs are the next gaming. Building is now more fun than playing a game.
This will equate to more time erosion for gaming in general... akin to TikTok et el.
I've realized I often would prefer to vibe code little parts of games than actually play games these days.
Especially cerebral ones that require some thinking like League.
I'm not quite sure what this really means for games but I think it's important?
Demand for far more powerful desktop PCs is about to explode. Power users will spend $50,000 to $100,000 on their own compute nodes: multi-GPU/NPU workstations with large memory pools, built to run frontier-class models and the agents on top of them. Within a few years, this will be normal among the roughly 200 million power users of intelligence worldwide.
The machine on your desk spends most of its life idle. Frontier models, local and cloud, raise its utilization by an order of magnitude. They plan, execute, and check their own work for hours without a human in the loop. Every step of that loop burns compute, and more of it will run locally.
The clearest example is professional software driven headlessly. An agent running Blender with no GUI can render, simulate, and iterate on a scene all night. The same holds for video encoding, code compilation, scientific modeling, and local inference. When agents drive these workloads around the clock, the GPU, NPU, and CPU carry more sustained load than at any point in the history of personal computing.
@KenneyNL This is bullshit, Kenney. I’m the same tech tree following optimist who moved from C to C++, from 2D to 3D, and from closed source to open source. I run a company. I write code. I contribute to published papers. I’ve upped my expectations of AI. Up yours!
there’s a ridiculous trend going viral on tiktok right now called “malfoid” a harry potter fanfic story where draco malfoy is a girl and in an enemies-to-lovers story line with harry
the AI videos are so good and it’s actively shifting AI sentiment on tiktok to positive (was historically super negative)
AI will evolve games and game dev.
It will NOT transform the games business.
There is substantial delusion about a "gaming" renaissance. This is not happening.
For those unaware... the console gaming biz has been flat for almost 20 years and is now declining. The overall TAM for gaming is contracting not growing.
Time share for games will continue to erode.
I think gen AI has a good chance of saving VR. Arguably it’s been content limited. And too hard to port existing games (where the market is). But many of those hurdles are now gone …
🫡
"And no, that's not the position of the Linux kernel.
I realize that some people really dislike AI, but this is an area where I'm willing to absolutely put my foot down as the top-level maintainer.
Linux is not one of those anti-AI projects, and if somebody has issues with that, they can do the open-source thing and fork it.
Or just walk away." - Linus T