Research and Product at @grayscale Prev VC at https://xitter.cf/t.co/JUtkZ0di2A | https://xitter.cf/t.co/PbetdAppFO Opinions are my own. Disclosure: https://xitter.cf/t.co/d1OPIuHW90
Joined May 2015
- Tweets961
- Following2.6K
- Followers1.6K
- Likes6.9K
Pinned Tweet
Nvidia H100s cost $50K+ to buy and can be rented on platforms like @akashnet for as little as $2 per hour. Hoping they can scale up supply.
Will Ogden Moore retweeted
AI compute infrastructure stocks ($CRW, $NBIS) have taken a beating of late. But compute demand continues to grow.
This is an opportunity to add exposure to the AI compute layer.
Read more on The Stack: grayscale.com/the-stack/comp…
Will Ogden Moore retweeted
The AI compute shortage is getting WORSE and what's happening with NVIDIA's B200 GPUs right now is absolutely INSANE.
According to Ornn Exchange, rental prices for NVIDIA's B200 GPUs just reached a new all time high of $8.54 per GPU hour, with prices climbing another 12.9% in just the past trading week. Normally, as technology gets older and more capacity comes online, you would expect rental prices to gradually decline. But we're seeing the exact opposite happening with B200s. Rental prices are moving higher, which shows that customers are willing to pay a premium just to secure access to compute.
This is an incredibly BULLISH signal for the economics of AI infrastructure because higher utilization and stronger rental pricing can translate into better returns on deployed GPUs, especially for operators that have already secured capacity. And this is exactly why I remain so bullish on the long term opportunity for neoclouds. These companies are spending billions building out AI infrastructure and one of the biggest concerns investors have had is whether all this capacity will eventually lead to oversupply and collapsing rental prices but right now, the data is telling us a VERY different story. And demand for compute is already this strong TODAY, imagine what happens as these technologies begin scaling across the global economy.
Yesterday, @OrnnExchange's B200 and B300 indexes reached a new all-time high prices, with the B200 hitting $8.54/GPU-hour.
It is clear that today's market is lacking GPU capacity, and that the companies building deploying capacity are looking to improve their credit. In the past trading week as the B200 rental price increased by 12.9%, $CRWV fell by 25.2%.
Why does $CRWV move inversely to rising compute prices as a neocloud? 98% of Coreweave's Q2 revenue came from committed contracts, 72% from three customers, and most of its $104B backlog is recognized after mid-2028.
CoreWeave's backlog is locked at fixed prices. These higher market rates matter for new contracts they sign, so the question is where rental rates end up when their contracts roll off.
Ornn's index tracks transacted prices in order to capture what buyers are actually willing to pay for on-demand GPU rentals, real signal for tracking the value of compute.
NEAR’s Q3 update highlights growth in Intents volume and revenue, private transactions becoming the default, NEAR AI Cloud’s OpenRouter integration for confidential inference, and plans to bring IronClaw into near.com.
Will Ogden Moore retweeted
Scarce electricity and power bottlenecks make Bitcoin’s hashrate loss a durable secular trend. BTC has lost its monopoly on the marginal megawatt, which now goes to inference (and Pearl)
Great chart from @hashrateindex comparing revenue per MWh across ZEC and BTC mining and AI compute.
Curious what H100 numbers would look like with additional revenue from PRL.
45% of NVIDIA data center revenue comes from neoclouds.
The largest company in the world has a strong incentive to make neoclouds succeed.
Will Ogden Moore retweeted
Neocloud stocks got absolutely crushed in the last month while $NVDA B300 GPU rental rates actually jumped more than 20% over past month.
The market is selling $NBIS, $IREN & $CRWV over concerns that GPUs lose value too quickly even as demand for open model inference is pushing rental prices higher.
🦪
Great chart from @hashrateindex comparing revenue per MWh across ZEC and BTC mining and AI compute.
Curious what H100 numbers would look like with additional revenue from PRL.
No better way to start a Sunday. All my homies love @jonbma and his bball runs.
The fundamentals haven’t changed: compute remains supply constrained, and GPU rental prices continue to rise as they have all year.
David Sacks pushes back on the AI bubble narrative and says the Corweave crash is just volatility
"I would make the argument that this is just volatility, which you're going to get a lot of whenever you're underwriting to multi-trillion dollar outcomes in the future based on very rapid, exponential rates of growth."
"The market is going to be super sensitive to any data points we have in the present. And so everything's gonna be exaggerated. And that's why you get these big swings."
"That to me is the explanation of volatility. I don't think that something fundamental has changed. Has the demand changed? No. We know that the demand for compute just keeps growing by leaps and bounds. In fact, everyone is saying that they are compute constrained."
"I mean, you got the biggest model companies are saying their compute constrained. Startups are saying it. I mean they're asking their VCs, hey, can you help us get compute. I think you've got some VCs even trying to buy compute as basically a strategy for competitive differentiation so they can help their startups."
"Old GPUs, like even the H100s or the A100s, which are several generations old, they're still renting at premium prices right now. They were supposed to be fully depreciated, and they're not."
There’s still a massive amount of compute capacity that could come online over the next few years, creating significant revenue growth potential for neoclouds and data center operators.
research.grayscale.com/repor…
Great chart from @hashrateindex comparing revenue per MWh across ZEC and BTC mining and AI compute.
Curious what H100 numbers would look like with additional revenue from PRL.
Will Ogden Moore retweeted
The value of human-powered information networks in the age of AI. Our core thesis 👇
longholdco.com/letters/human…
AI is redefining the value of human-to-human information flows.
For 600 years, value for media and information companies came from aggregating distribution and audiences. First by owning presses and trucks, then by renting reach from search and social.
AI breaks that. Anything already online can be summarized and piped through an API.
What stays scarce is information that only moves between people: what an executive tells a reporter he trusts, what a buyer tells a seller she’s known for ten years, what's said at a hosted-buyer event.
In this new era, value accrues to whoever best aligns these human-to-human information networks. This is one of our core theses at LongHold.
People need to recognize that AI is inevitable. There’s always been resistance to change with new tech. But the stakes here are different. There is a range of outcomes, good and bad, for the majority of society. Across builders/investors/policy, more people need to communicate the positive potential outcomes, explain how this could benefit ordinary people, and build toward those outcomes to prevent the negative outcomes from occurring.
Every person should be able to gain access to the private markets.
It’s great to see the SEC proposing merit-based private market access in the US. This will be a huge step forward.
For too long, only the richest people have had access to the private markets. Retail has been missing out.
And over the last few years, companies have been incentivized to stay private longer (so retail has been missing out even more at best, or become exit liquidity at worst).
This is the future. Open weight models fine tuned for a company’s specific internal needs. Faster, more cost effective, and will be run locally particularly in cases where privacy is paramount.
We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case).
I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves.
We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them:
"/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started"
I imagine we'll have some backend services that could be replaced entirely by this new architecture.
