The alpha, twice a week
That was 5 min.
Get the best new moments in your inbox every Wednesday and Saturday. It's free.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
Moonshots · 5 min · Nov 16, 2026
Compute bids turned against Figure as GPU supply tightened
Brett Adcock says Figure spends low nine figures annually on compute and expects billion-dollar Helix training runs next summer.
In this episode
Full analysis- 16:19China's robot dance demos do not prove practical autonomyPro pick 3 min
- 24:06Figure runs its robot intelligence on board to meet latency needs5 min
- 1:01:00Figure expects lights-out robot manufacturing within five years4 min
- 3:19Figure will wait for a high-quality home robot before launch6 min
- 12:18Figure’s coffee-cup demo helped validate neural nets in robotics4 min
- 36:48Figure built its own actuator after a supplier’s part was too large3 min
- 1:12:11Figure estimates Baku capacity at over 40,000 robots a year3 min
- 1:43:27Compute bids turned against Figure as GPU supply tightened5 min
“I was like, in 2027, you're not going to be able to compute anymore. And that happened in, like, September of 2026 for us.”
Brett Adcockat 1:45:21 · MoonshotsTakeaways
- Figure spends low nine figures a year on compute.
- Adcock expects billion-dollar Helix training runs next summer.
- By September 2026, bids for additional GPU capacity had dried up.
About this segment
Brett Adcock says Figure spends low nine figures annually on compute and expects billion-dollar Helix training runs next summer. When it sought more capacity, GPU bids became scarce and more expensive.
Brett Adcock is founder and CEO of Figure AI.
Founded Figure AI to build general-purpose humanoid robots; previously co-founded Archer Aviation and Vettery.

Peter Diamandis, Host · Founder, XPRIZE
Co-founded Singularity University and founded Zero-G; co-wrote the books Abundance and Bold.

Dave Blundin, Host · Founder & GP, Link Ventures
Has co-founded 23 companies; founding CEO of Vestmark and chairman of insurance marketplace EverQuote.
From “Robotic Deep-Dive: Beating China, the Race Against Tesla, & New Robotics Timeline With Brett Adcock”, Nov 16, 2026. Watch the full episode
Answers “Is there still a GPU shortage?” See all 47 answers
Thanks. We'll look at and fix what's off.
Transcript
Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.
- 1:43:31
over the last like month, more than half of our compute has been taken from like one giant pre-training run. A large percentage of it. So like... What's the price tag on a big, big run for you? I mean, we pay like, and we pay nine figures a year for compute right now. Like low nine figures. So like, so like, you know, that was like maybe 50 to 60, 70% of our compute will be used for pre-training, so a lot, nine figures a year, of just going into pre-training for Helix, Helix 3.
- 1:44:03
So quite a lot. If we had 10 times more compute, we would train on 10 times more compute. Well, next summer you have 100 times more, are you going to do a billion-dollar training run? Yeah. I think, hopefully, I expect that next summer we'll be doing billion-dollar training runs on Helix. If all goes well, it's like everything's got to go well, but it looks like it. It looks like you'll have the data, enough data.
- 1:44:24
We'll have the data. We have the compute, we have the capital. Getting these larger clusters set up is hard. So we went out six, seven months ago for compute. It's the current cluster we have now. I think it was seven months ago. We had four vendors that came back that had compute, pretty decent chips, new Blackwell series chips. And we beat the piss out of them in pricing. Just got it down. The GPU per hour, it was a great deal.
- 1:44:50
We signed it, great, it's going on. We went out, I think it was three months ago, four months ago, a couple months after that. We were like, we need to double our compute capacity. We were just needing more for Helix 3. We went out, and we got like... We went out and got no bids back. And then I was like, we've got to push harder. We went out again. We got one bid back from a big cluster. And then they beat the piss out of us on price.
- 1:45:11
Like buyers and sellers have flipped. I was like, what just happened? And we just couldn't get any more compute. And we met the NVIDIA team. And I was like, listen, it's just where everything's at capacity. The whole world is just there. And I made this prediction earlier in the year. I was like, in 2027, you're not going to be able to compute anymore. And that happened in, like, September of 2026 for us.
- 1:45:32
September 2026, the world ran out of compute. You just can't get it. I'm so glad you're saying that. Because so many people are misinformed right now about, oh, you know, the SAI thing is overblown. The bubble is going to be chips lying everywhere. We can't get H100s. We can't get the old chips. They're trading above par. We can't get those. Somebody wrote me yesterday, like, I got compute. I'm like, great.
- 1:45:51
I'm out. She's like, eight nodes. I'm like, dude, eight nodes. Like, I can't even stop and talk to your eight nodes. It's like, I need more than eight nodes, man. It's not going to work at all. So, yeah, the whole world's out of compute. And then compute is constrained by memory. The whole world's out of memory. It takes three or four years to build those fabs from scratch. So, yeah, there's like a really...
- 1:46:11
And NVIDIA has been like one of the best places. You can't train on anything else besides this. So, it's really just interesting environment. You know what else is, you mentioned staying in the private markets longer, but a lot of the investment is coming from NVIDIA and other public companies. So it's actually coming from the public markets through large public companies, but they never used to invest in startups or early stage companies or even like fast growing companies like FIGURE.
- 1:46:35
Now they're just wide open for business. That's also a Silicon Valley kind of invention, which allows much bigger things to be built highly efficiently in the private market. Maybe a fab, you know, that would take it to the next level. Because you were saying a second ago, like the new fab, the new HBM memory fabs won't be online for like three years. but if you look at how many are being built it'll be over overbought before they even come online yeah like it's not even even if they come out on time which they won't like none of these are forecasting humanoids yeah they're not even forecasting not even tonight i mean like we literally met the cto of one of the big groups and they're like we don't even have any of these center projections but you know everywhere you go to whether it's software for mechanical design or
- 1:47:14
whether it's space whatever everyone's like yeah they haven't forecasted my demand yet yeah like okay we have enough of these conversations you stitch it all together you're like oh my god we're like i mean if you look at like the scaling laws for compute too like um literally just the amount of data and same parameter models being trained on more compute It has like log linear relationship. You're seeing like a real ROI there.
- 1:47:36
So I would say like, man, it doesn't seem like this trend is anywhere close to stopping. So it just, I think we're really early in the, what's coming out of this, like both digital systems, like, you know, say we're doing a Hark or even like figure stuff, like it's just so magical. It's like, it's like going to be way, it's going to be a hundred times bigger than the It's like this stuff is going to be so important.
- 1:47:58
A thousand times bigger. So I think we're just like, I don't know. My view is we're still really early, and it's not time to panic yet, so I keep calm and keep building. So I'm a Hark Pro user. Nice.