StarZero All shows Try the podcast clipper →

George Hotz

Founder, Tiny Corp · 1 appearance on the record.

computed by AI from the episodes · how this works → · full disclaimer →

founderengineer@realgeorgehotz ↗geohot.com ↗Wikipedia ↗Instagram ↗

George Hotz first gained notoriety for developing the first carrier unlock for the original iPhone and reverse-engineering the PlayStation 3. He subsequently founded comma.ai to build open-source autonomous driving systems and later launched the tiny corp to build the deep-learning framework tinygrad.

32statements → 16claims → 9claims resolved → 67%fully supported → 3.88/5average certainty → 2.94/5average debate potential → ≈4.0/5argument clarity, estimated → 25said about them ↓

6 supported 1 partly supported 2 contradicted 2 not yet assessed 5 not checkable as stated how the 16 claims stand · each chip opens the sources

5 predictions · 11 assertions · 8 opinions · 5 insights · 3 disclosures · every statement was checked. The predictions and assertions are the 16 claims: statements the public record can support or contradict. 9 are resolved, 2 are not yet assessed, and 5 name no date, number or outcome precise enough to check. Everything else (opinions, insights, what ifs, disclosures) can never be settled by the record, so it carries no assessment.

The record, in short

What the tape says about how George argues and how the claims held up. Everything they said, and everything said about them, is in the tabs below.

Their most notable supported claim

Assertion Supported
Hotz: Tinygrad runs all ML models with only 25 primitive operations
“Tiny grad is, we are going to make a risk offset for all ML models. And yeah, it can run all ML models with basically 25 instead of the two 50 of XLA or PrimTorch. So about 10 X less complex.”
George Hotz Jun 20, 2023 ▶ 3:51 Ep 18: Petaflops to the People — with George Hotz of tinycorp

Their most notable contradicted claim

Assertion Contradicted
Hotz: Nobody has successfully trained models in INT8
“No one's gotten training to work with Indate yet. There's a few papers that vaguely show it, but if you're training, you're going to need BF-sixteen or float-sixteen.”
George Hotz Jun 20, 2023 ▶ 34:48 Ep 18: Petaflops to the People — with George Hotz of tinycorp

Expressed certainty vs assessment result

none yet certainty 1
none yet certainty 2
none yet certainty 3
69% certainty 4
100% certainty 5

weighted support: a fully supported claim counts one, a partly supported claim counts half. Each filled bar is clickable and opens exactly those claims; "none yet" means nothing said at that certainty level has resolved yet

How they sound: not measured why? →

We measure speaking style by listening to the audio itself, and a fair number needs at least 2,000 words from one person on tape we have measured. There is too little of George Hotz on measured tape to publish a rate. This says nothing about how they speak.

Everything George Hotz said on Latent Space that made the record, most notable first. Filter by type, assessment or year in the ledger →

Assertion Not publicly verifiable
Hotz: GPT-4 is an 8-way mixture model with 220B parameters per head
“GPT-IV is two hundred twenty billion in each head, and then it's an eight-way mixture model.”
George Hotz Jun 20, 2023 ▶ 49:48 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Opinion
Hotz: Meta attracts researchers who want to publish while OpenAI keeps ideologues
“OpenAI can keep ideologues who, you know, believe ideological stuff, and Facebook can keep every researcher who's like, dude, I just want to build AI and publish it.”
George Hotz Jun 20, 2023 ▶ 55:16 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Opinion
Hotz: Analog computing for AI won't work and clockless chips aren't practical
“Analog computing just won't work. And clockless computing sure, it might work in theory, but your ETA tools are, maybe AIs will be able to design clockless chips, but not humans.”
George Hotz Jun 20, 2023 ▶ 7:50 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Opinion
Hotz: Systolic arrays are the wrong architectural choice for AI chips
“I think systolic arrays are the wrong choice. Systolic array, I think they have systolic arrays because that was the guy's PhD. And of course Amazon makes... They are very power efficient, but it becomes hard to schedule a lot of stuff. On them, if you're not …”
George Hotz Jun 20, 2023 ▶ 9:12 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Prediction Not checkable as stated
Hotz: Tinybox will be 5x faster than H100 systems per dollar
“For 90% of most companies model training use cases, the tiny box will be five X faster for the same price.”
George Hotz Jun 20, 2023 ▶ 48:45 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Prediction Not checkable as stated
Hotz: Machines will replace all human labor in about 20 years
“I'm a believer that machines are gonna replace everything in about 20 years.”
George Hotz Jun 20, 2023 ▶ 55:44 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Insight
Hotz: Maximally compressed human brain represents only a couple gigabytes
“Quantization is a poor man's compression. I think we're only talking really here about, like, maybe a couple gigabytes, right? And then if you have, like, a couple gigabytes of true information of yourself up there, cool man. Like, what does it mean for me to …”
George Hotz Jun 20, 2023 ▶ 1:10:05 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Insight
Hotz: Real AI alignment problem is corporate and government misalignment
“I think it's actually not a question of whether the computer is aligned with the company who owns the computer. It's a question of whether that company's aligned with you or that government's aligned with you. And the answer is no. And that's how you end up de…”
George Hotz Jun 20, 2023 ▶ 1:11:38 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Opinion
Hotz: Qualcomm makes the best inference chips
“Qualcomm has the best inference chips.”
George Hotz Jun 20, 2023 ▶ 2:47 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Assertion Supported
Hotz: Tinygrad runs all ML models with only 25 primitive operations
“Tiny grad is, we are going to make a risk offset for all ML models. And yeah, it can run all ML models with basically 25 instead of the two 50 of XLA or PrimTorch. So about 10 X less complex.”
George Hotz Jun 20, 2023 ▶ 3:51 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Opinion
Hotz: Google TPUs are the only successful non-Nvidia training chips
“The only company, there's one other company aside from Nvidia who's succeeded at all at making training chips... Mid journey is trained on TPU, right? Like a lot of startups do actually train on TPUs, and they're the only other successful training chip aside f…”
George Hotz Jun 20, 2023 ▶ 4:33 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Disclosure
Hotz: AMD CEO Lisa Su sent pre-release ROCm 5.6 to fix panics
“Lisa Sue reached out, connected with a whole bunch of different people. They sent me a pre-release version of Rock M 5.6. They told me you can't release it, which I'm like, okay, Why do you care? But they say they're going to release it by the end of the month…”
George Hotz Jun 20, 2023 ▶ 26:02 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Prediction Not checkable as stated
Hotz: Best chatbots will be smaller models with 1,000 training runs
“I don't think that the best chatbot models are going to be the big ones. I think the best chatbot models are going to be the ones where you had a thousand training runs instead of one. And I don't think that the interconnect bandwidth is going to matter that m…”
George Hotz Jun 20, 2023 ▶ 42:48 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Disclosure
Hotz: Third company will build an AI girlfriend product
“The third company's the first one that's gonna build a real product, and that product is A girlfriend? No, like I'm dead serious, right? Like this is the dream product, right? This is the absolute dream product.”
George Hotz Jun 20, 2023 ▶ 1:07:36 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Opinion
Hotz: Merging with machines requires an AI companion, not Neuralink electrodes
“So I don't need to put, you know, electrodes in my brain to merge with a machine. I need an AI girlfriend, right?”
George Hotz Jun 20, 2023 ▶ 1:08:23 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Disclosure
Hotz: tinycorp will eventually partner or manufacture its own chips
“So I'd like to start another organization that eventually in the limit either works with people to make chips or makes chips itself and makes them available to anybody.”
George Hotz Jun 20, 2023 ▶ 2:49 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Assertion Supported
Hotz: Tinygrad runs OpenPilot in production 2x faster than Qualcomm's library
“TinyGrad is used to run the model in OpenPilot. Like right now, it's been live in production now for six months. And TinyGrad is about two X faster on the GPU than Qualcomm's library.”
George Hotz Jun 20, 2023 ▶ 21:36 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Prediction Not checkable as stated
Hotz: Tinygrad could replicate PyTorch's API in two engineer-months
“Replicating the PyTorch API. Is something I can do with a couple, you know, like an engineer month or two.”
George Hotz Jun 20, 2023 ▶ 31:32 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Assertion Contradicted
Hotz: Nobody has successfully trained models in INT8
“No one's gotten training to work with Indate yet. There's a few papers that vaguely show it, but if you're training, you're going to need BF-sixteen or float-sixteen.”
George Hotz Jun 20, 2023 ▶ 34:48 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Assertion Not checkable as stated
Hotz: None of 20 tested PCIe extenders work at PCIe 4.0
“No PCI extender I've tested and I've bought 20 of them works at PCIe four point out. So you're going to need PCIe redrivers now.”
George Hotz Jun 20, 2023 ▶ 38:57 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Insight
Hotz: Except for Apple, secretive tech companies hide underwhelming tech
“Whenever a company is secretive, with the exception of Apple, Apple's the only exception, whenever a company is secretive, it's because they're hiding something that's not that cool.”
George Hotz Jun 20, 2023 ▶ 50:37 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Insight
Hotz: Training on internet cross-entropy loss yields mediocre AI responses
“The problem is, if your loss function is categorical across entropy on the internet, your responses will always be mid.”
George Hotz Jun 20, 2023 ▶ 57:15 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Opinion
Hotz: RLHF models adopt customer support personalities
“I don't like the RLHF models. I don't like the tuned versions of them. I think that they become, you take on the personality of a customer support agent, right?”
George Hotz Jun 20, 2023 ▶ 1:06:11 Ep 18: Petaflops to the People — with George Hotz of tinycorp
Opinion
Hotz: Nvidia makes the best training chips
“NVIDIA has the best training chips.”
George Hotz Jun 20, 2023 ▶ 2:46 Ep 18: Petaflops to the People — with George Hotz of tinycorp

Show 8statements(8 left)

The other half of the tape: George Hotz's own voice is left out of every number here. Other people bring the name up 24 times in 14 episodes on Latent Space. 1 statement on the record names them. every mention, with the transcript →

Who brings them up most Shawn Wang 9Eric Simons 3Alessio Fanelli 1

Statements about George Hotz, by other people (1)

Prediction Not checkable as stated
Chintala: George Hotz's TinyGrad requires major breakthroughs to match PyTorch
“There's no, like, I don't think, like, unless we have, like, great breakthroughs, like, George's vision is achievable, like, or, like, he should be thinking about a narrower problem, such as, I'm only gonna make this for, like, work for self-driving car con ne…”
Soumith Chintala Mar 6, 2024 ▶ 9:40 Open Source AI is AI we can Trust — with Soumith Chintala of Meta AI

Every mention by year

tap a year for its mentions
0083155202320242025episodesmentions
035202320242025episodes it came up in
001.32.52.55202320242025episodesmentions per episode
2025 6 mentions in 4 episodes 2 per episode
2024 11 mentions in 5 episodes 2 per episode
2023 7 mentions in 5 episodes 1 per episode

Appearances (1)

EpisodeDateSpeaking time
Ep 18: Petaflops to the People — with George Hotz of tinycorp Jun 20, 2023 59m
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 200 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more