i joined Nvidia two years ago, and there are a few things i wanted to say.
one of the biggest surprises after joining was finding out Jensen actually read employee emails. he'd write back with informed feedback and sharp, to-the-point questions.
somewhere along the way i decided that presenting in a meeting with Jensen in the room would be an awesome career milestone. i only let myself aim for it because he was that hands-on.
it hurts to know that dream of mine is dead now, and i'm deeply disappointed by how this whole thing has played out.
my fascination with NV goes back much further than that. as a kid i begged my parents for a decent GPU so i could play CoD. and 10 years back, when i wanted to get into AI, i saved up for a 1080 so i could run CUDA. i've followed Jensen closely ever since, watched every interview, and my respect for him only grew. to me, he's a one-of-one visionary and one of the most consequential entrepreneurs of all time.
Jensen stood for something bigger than his own talents: open models. i have a hard time imagining whoever comes next fighting for open models the way he did, since i believe that's exactly why he was removed.
we're at one of the most consequential inflection points in history. the stakes are even higher than the original open-source movement, and fundamentally different: this time, open models won't win by default—we have to fight to keep open source alive. if we don't, we'll pay for it in privacy, security, and extreme power concentration.
Jensen was leading that fight, and he was perhaps the only one who could.
You can learn a lot about a university's values by noticing when it protects someone from the mob and heckler's veto and when it just...lets things happen.
You can also learn a lot by noticing who behaves this way on campus when they don't get their way and...who doesn't.
@gwenshap yeah, it's working
powers mj's notifications for example, and half a dozen other places it's useful to add an ai-powered if statement
(well, we switched to GLiDE from Fastino, but same family)
not really a flashy thing, just useful
https://t.co/ODoxQLRxub
This is the Turing's roadmap for the UK AI sovereign model. I'm a little confused about whether this is a plan or a plan to make a plan but as far as I can tell they're quite serious about it
Text editors are a lot of fun to work on because you'll select a bunch of text and the browser will totally lock up so you profile it and you learn that Apple's AI Writing Tools run on all of the selected text, every time your mouse moves.
There is a common belief in the AI field—that advanced AI means the inevitable forfeiture of human primacy in the world. Even many who disfavor that future believe it is likely. I call this the silent consensus, and it must not be allowed to remain silent any longer.
It's depressing if you think about how much Google spent to build this harness that is much worse than just about any open source alternative would have been.
Google Antigravity just turned 1 and reset usage quota for everyone across all models. Here's what you need to know.
Varun Mohan posted on Nov 18, 2026: "One more thing... In honor of Antigravity turning 1, we're resetting quota for everyone across all models!" The post pulled about 128K views. Users confirmed their usage meters reset within hours.
Antigravity launched on Nov 18, 2025, alongside Gemini 3 Pro. It was an agent-first IDE with browser control to test apps automatically, and an Agent Manager to run parallel agents. Kevin Hou, who helped launch it, says it's now the norm to run hundreds of agents at once.
This follows the Gemini 4 Hydrogen and Antigravity robotics news we covered earlier today. Mohan also said time to first token for Carbon is now 1.5x faster on Antigravity, with Argon to follow soon. Time to first token is how long you wait after hitting send before the model starts replying.
User reaction was mixed. Some were glad to get a reset, including one user who was at 20% of their weekly limit. Others had hoped to be warned first. One user said the reset came with no warning while they were at 90% of their weekly limit, and they asked for advance notice like OpenAI's Codex resets. Another user noted that some models aren't available on free or Pro plans.
Key numbers:
- Launched Nov 18, 2025, so now 1 year old
- Quota reset applies to all users and all models
- Carbon time to first token is 1.5x faster
- Argon speedup is promised next
Kevin Hou pointed users to features like /swarm, /browser, remote agents, Carbon, Hydrogen and Gemini Orchestrator.
im so confused by gemini argon and carbon. they're the same size (same price and latency) but carbon is much better than argon. why would anyone use argon? why even release it? is argon better at anything at all? what is going on.
Wait until Harvard resarchers will find 12-18 months later that the bottleneck shifted from code review to CI/CD (first) and then to code verification + o11y
Obituaries
The demise of Jensen Huang was the end of a great many things, including Nvidia’s long-term prospects, the AI bubble, and the exceptional way in which the U.S. viewed itself in the world.
https://t.co/QT0XIxFwXJ
If you ignored every unit of housing, bridge, beach or mile of road ... Robert Moses still built 650 playgrounds in NYC in 26 years
Thats a new playground every other week for 2 1/2 decades
It’s an unfathomable level of competence for a public servant today. It cannot be done
Feels a little weird that OpenAI published this chart about their internal model and then hasn't said anything about where GPT 6.2 Astra falls into this picture?
And 15% only gets you a meeting. Removing directors without cause takes 2/3rds of all outstanding shares, not just the shares that show up to vote. Removal for cause only takes a majority of all outstanding shares. But a board enforcing the co's own code of conduct isn’t cause.
Part of this is because we don't nerf Carbon for real world frontier research in coding, math, AI, science etc.
The ability to push the frontier is open to all with Carbon :). Go forth and discover.
Hearing the @united CEO Scott Kirby on @CheekyPint telling @collision that the primary constraint on commercial aviation in the US is ATC slots drives me crazy. There are at most about 5000 flights occurring at any one time over the entire US. There are about 30 usable flight levels in alpha airspace, and you can deconflict 100x more flights than this with a single toothbrush CPU (https://t.co/1BdhZ1Ap58).
In terms of physical space, there's enough room to fly every man woman and child in the US simultaneously. Look up. Does the sky seem crowded?
There were plans to automate ATC using the first Iridium constellation more than 30 years ago.
At some point we have to have a serious conversation about rent takings by obsolete incumbents that are blocking entire percentage points of economic growth.