Jonathan Yang

@NorthwesternU | prev @NVIDIA, @PalantirTech

[Jonathan Yang @Jonathann_Yang]

@saralucijad This would be so clutch 🙏🙏🙏

[Jonathan Yang @Jonathann_Yang]

@sebonomics @alexandr_wang @chameleon_jeff @ScottWu46 @aryansfv @jakeschuller6

[Jonathan Yang @Jonathann_Yang]

Wow congrats Tyler!!

[Tyler Fong @tyler_fong_]

Today I’m joining @Natl_Compute

I’ve spent the last four years working on GPU access, first as a founding engineer at Brev and then at NVIDIA after they acquired us in 2024. When I started, on-demand GPU capacity at the hyperscalers was effectively gone. If you were an independent researcher, a two-person startup, or a grad student with a grant and a deadline, you refreshed console pages and got nothing, or hammered the API only to get “insufficient capacity” errors back. I vividly remember hard-coding a filter at Brev to hide certain GCP instances because they almost never provisioned.

Around 2023, a wave of new public clouds were coming online, and many of them had on-demand capacity. That gap is a big part of why Brev worked: we gave researchers a simple way to find GPUs across dozens of providers and start building. For a while, aggregation was enough.

Then in 2025, I watched that capacity dry up too.
Access to compute has never been worse. Existing supply is already under contract, and new capacity is reserved before it ever comes online. I won’t rehash why. There are a hundred good explanations out there covering debt financing, anchor tenants, backstops, lead times and so on.

What matters more are the downstream effects. Long-term, single-tenant leases hold back scaling for everyone who couldn’t secure funding in time, and the industry’s momentum shouldn’t hinge on the handful of players who locked up access early.

Today small startups, labs, and university researchers don’t even get a chance to compete. National Compute is solving this with a foundational layer we call “the Grid”. It’s an interconnected network of multi-silicon capacity composed of multiple providers and spans multiple sites that will be pooled for our members.

The speed at which this team has been able to scale in such a small fraction of time is remarkable, and I’m excited to push that scale even higher for the public benefit.

[Jonathan Yang @Jonathann_Yang]

@NVIDIARTXSpark Hey @Baxate

[Jonathan Yang @Jonathann_Yang]

@NaderLikeLadder Alec is locked in

[Jonathan Yang @Jonathann_Yang]

@tyler_fong_ @nvidia Tyler!! Absolute 🐐

[Jonathan Yang @Jonathann_Yang]

Griffin too. The latency is unbelievable

[Alex Finn @AlexFinn]

We are living in the greatest 30 day time period in technology history

In the last month all of these have been released/are expected to be released:

• Opus 5.5: greatest AI model ever
• Meta Muse: The AI agent that will get a billion new people into AI agents
• ChatGPT 6 Astra: First model that could be considered AGI
• Grok 4.7: affordable version of Opus + Elon’s back
• GPT 6.1 sol: Basically free intelligence
• Sonnet 5.5 Dirt cheap intelligence you basically can't tell apart from Opus
• ChatGPT Dot: Incredibly powerful AI agent that's capable of building out large scale applications

The way you work 30 days from now will be dramatically different than the way you work today

You will be significantly more productive with significantly less effort required

Thank your God he/she decided you should be alive right now

[Jonathan Yang @Jonathann_Yang]

GPUs go brr

[amit @amitisinvesting]

THE US GOVERNMENT LAUNCHES AMERICA DOT GOV, A CUSTOM LARGE LANGUAGE MODEL BUILT TO ANSWER QUESTIONS FOR 40M DAILY USERS ACROSS 29,000 GOVERNMENT WEBSITES.

Yes, it's a chatbot, but to be honest...if the entire world is now engaging with LLMs, it was time for the US gov to transform how citizens access information.

It was also announced that the website will go beyond information and help you take action starting next year, exciting to see some new usecases for this technology in the public sector.

[Jonathan Yang @Jonathann_Yang]

We will have fable level intelligence that can run locally on a spark/studio soon

[Ahmad @TheAhmadOsman]

234 days later, we got Opus 4.5 / 4.6 intelligence and capabilities running on a single RTX 3060 with 8GB VRAM

[Jonathan Yang @Jonathann_Yang]

@AsherStrell Congrats, do you want a cookie?

[Jonathan Yang @Jonathann_Yang]

beautiful

[Palantir @PalantirTech]

“Dr. Karp talks about chips and ontology. This is ontology for chips.”

At AIPCon 11, @nvidia reveals how it is partnering with Palantir to build and run their supply chain on Palantir.

[Jonathan Yang @Jonathann_Yang]

Probably tbh… Supercell took down the post so you’re safe Ken 🤝

[Ken @kfcclash]

they'll replace the surgeons before me

[Jonathan Yang @Jonathann_Yang]

@ryanirl How did you post train this model

[Jonathan Yang @Jonathann_Yang]

@noobeycr Not me, that’s even cooler tbh I think that person has their own CV layer + model

[Jonathan Yang @Jonathann_Yang]

Astra’s hidden talent: playing Clash Royale 😭😭

I installed clash via blue stacks so computer use would be able to interact with the game. It’s currently on a 5 win streak (it’s a new account tho, so the opponents are pretty shit)

Latency is definitely the biggest issue right now - around 1-2 seconds from opponent acting to gpt registering it.

This is cool. Will keep playing to get to higher trophy opponents @kfcclash watch out

[Jonathan Yang @Jonathann_Yang]

@sebonomics Power will be the next bottleneck + investment wave. I give it 3 months and everyone will be talking abt it

[Jonathan Yang @Jonathann_Yang]

@ZelrigX Sure, but all you need to do is

1) Download bluestacks (or some equivalent emulator)
2) Download clash royale on the emulator
3) Open up the ChatGPT app and tell it to use computer use to play a game

Lmk if that works

[Jonathan Yang @Jonathann_Yang]

@Andrew140904 Might give this a try

[Jonathan Yang @Jonathann_Yang]

@yummyslider Hahah fair

[Jonathan Yang @Jonathann_Yang]

@MrxFlip Yea good call - this is an alt acc tho so I think my main should be good

[Jonathan Yang @Jonathann_Yang]

@WiggerNilly Will do!