Founder and CEO @Hyperbolic_Labs. ex-@citsecurities. Finished Math PhD in 2yrs @UCBerkeley. Math Olympiad Gold Medalist. Highest honor @PKU1898
California, USA
Joined November 2018
- Tweets4.8K
- Following1.9K
- Followers15.4K
- Likes7.1K
Pinned Tweet
AI is great at hitting explicit goals, but often at the cost of the hidden ones.
Terence Tao just wrote about this. He points out: AI is the ultimate executor of Goodhart’s law, i.e. when a measure becomes the target, it stops measuring what we care about.
Take a call center. Management sets a KPI: “shorten average call time.” Sounds reasonable: shorter calls should mean faster resolutions, happier customers.
At first, it works. Agents become more efficient. But soon, people start gaming it: nudging customers to hang up when the problem is tricky, or just dropping the call themselves.
The numbers look amazing. Call times plummet. But customer satisfaction? Straight into the ground.
Now replace “call time” with “prove theorem X.”
If human mathematicians did it, they’d refine definitions, polish lemmas, contribute back to Mathlib, train juniors, deepen the understanding of math structures, and strengthen the community.
The AI, by contrast, optimizes only for the explicit goal. It might generate a 10,000-line proof in hours. Perfectly correct, but unreadable, unusable, and useless for human learning.
The summit is reached but the forest along the way is gone.
We need to start making our implicit goals explicit and design systems that protect the values we actually care about, not just the numbers we can measure.
GPUs are becoming money-printing machines.
Lock in B300s at $4–5/GPU-hour and sell the capacity at $10–20/hour.
That's 2–5x the rental price.
Neolabs are already hoarding GPUs.
If you're sitting on H100s, B200s, or B300s, Hyperbolic can help turn your GPU capacity into revenue.
Start hoarding compute and monetizing it.
There is an absolute barrage of inference providers popping up that get to $1M/mo extremely quickly.
This is the rough economics of inference providers:
Even if you rent on-demand B300s at $8/GPU-hr, many open source models (GLM 5.3, Kimi K3, DeepSeek 4.1 Flash / Pro) can get you >$10-20/GPU-hr at reasonable throughputs (100tok/s). DeepInfer from DeepSeek directly shows they can get $38/GPU-hr on off-peak hours. Every chip / model has a different throughput / interactivity curve, but there is a lot of arbitrage between even high rental spot prices and token prices. New chips (Vera Rubins) also change this meaningfully. Your ability to be making gross margin relies on your utilization and how good your stack is.
Gateways like OpenRouter, Vercel and more help solve that problem by aggregating demand to you. May inference providers discount their tokens to get longer term commits too. I've even heard stories of selling back Chinese model tokens to Chinese labs!
One thorn in the economics of these players is that many Chinese open source model labs are now asking for rev share if you're at a certain scale, espp for pre-release access to their models for day 1 support. That further pushes the threshold to be in the money.
The barrier to entry seems low and falling. There are even stories of smaller ones who can "vibe code" a way to stand up OSS models on a cluster they rent. And as a result, we're seeing 10s of such companies pop up almost every week and immediately skyrocketing revenue at low-ish gross margin and raising rounds on the back of that.
Unsurprisingly, because the actual service provided seems commodity, this eventually drives up compute prices and drives down token prices. In theory, it would seem like neoclouds who have the lowest cost of compute and themselves have started inference businesses would win the race to the bottom (Crusoe Coreweave Nebius and GMI all have inference businesses), with larger players (Modal Fireworks Baseten) being second because they are well capitalized to make bigger compute commits cheaply and have demand (high utilization), but the reality is the pie is growing so quickly that even upstarts are absolutely crushing it and growing aggressively (inference dot net, DeepInfra, Wafer, Relace, Morph).
Strategies vary across the board: neoclouds want to serve inference, inference players want to be neoclouds and do higher margin reserved instances / post training, and some want to go heterogeneous hardware or creatively flexible compute constructs.
The risks are: do open source models continue to be competitive? Will their creators rent seek more? Will their shape change compute needs? How do different chips change economics? Who gets compute in a supply crunch?
Below is a aggressive sketch of how gross margin scales with utilization for various models / chip sets.
For now, there are many inference providers and they’re all ripping!
Jasper retweeted
Berlin rooftop, cocktails, and great conversations about compute, models, and inference 🍸
Thanks to everyone who joined us for Cocktails & Compute at @NVIDIAGTC Berlin!
Which city should we host next?
Excited to launch Flux, the next generation of Hyperbolic inference! 🚀
Efficient inference requires both optimized model serving and smart orchestration. How you allocate GPUs, place workloads, and improve utilization directly shapes the cost and performance of every token.
With Flux built on Forge, we’re bringing model serving and compute orchestration together, working toward a future where inference workloads can dynamically use available capacity across our global compute grid.
Huge thanks to the team for bringing this to life across engineering, product, and marketing. Proud of what we’ve shipped and excited for what comes next.
Give Flux a try and let us know what you think!
Introducing Flux, Hyperbolic’s next generation of inference, built on Forge.
Choose from 10 open models through one OpenAI-compatible API.
Start building with Flux → app.hyperbolic.ai/models
Explore the models and architecture: hyperbolic.ai/blog/flux-infe…
Proud to partner with @EagleCorp as an official launch partner for EAGLE-4 - bringing their fastest spec decoding solution to @hyperbolic_labs and accelerating the future of AI!
Congrats to the @EagleCorp team on EAGLE-4, their fastest yet, hitting 961.6 tokens/s on Qwen3.8-27B!
Proud to be an official launch partner, with EAGLE-4 coming soon to Hyperbolic.
ALT Eagle and Hyperbolic logos, Launch Partner, Accelerating the future of AI inference.
Jasper retweeted
GPU costs can be locked in. Inference revenue can’t always be.
New research from IOSF, in collaboration with our team, explores the “crack spread of intelligence”: the gap between GPU costs and inference revenue.
AI compute demand is accelerating, and we’re seeing it firsthand at @hyperbolic_labs.
Great conversation with the @blockspace team about GPU demand, supply bottlenecks, and the future of the AI compute market. Thanks for having us!
Why $NVDA GPUs Sell Out In Hours
Blockspace sat down with @hyperbolic_labs. The SemiAnalysis rated neocloud is one of the fastest growing companies in compute.
We discuss:
• Velocity on B300 sales
• Bottlenecks for AI Labs & neoclouds
• Trends in compute sales from the front lines
Many neoclouds are essentially hardware leasing businesses, taking on massive debt and long-term commitments while having limited software differentiation. That creates significant risk when GPU prices decline.
This is exactly why we’re building Hyperbolic differently. Rather than betting on the value of hardware, we’re building the software and market infrastructure to unify fragmented compute supply, improve utilization, and provide liquidity across the global GPU market.
We believe the long-term winners won’t necessarily be those who own the most GPUs, but those who can most efficiently allocate, orchestrate, and monetize compute.
In fact, declining GPU prices could benefit platforms like ours by making compute more accessible and expanding overall demand.
Ben Horowitz explains why a16z doesn't invest in neoclouds after living through the dotcom bubble
"If you get into a pathological debt situation, that can wreck anybody. It's why, by the way, we're not invested in any of these neoclouds, for that reason. That whole thing scares me, having been through it."
"It can end up being a very dangerous arbitrage business where you can get upside down easily."
"You look at Loud Cloud just as an example, here's a simple thing that happened to us in 1999. Bandwidth was $1,500 a megabit. And you couldn't get it, it was very hard to get. And so you had to get a long-term contract, three years."
"By 2001 it was $50 a megabit. So if you were holding bandwidth, you were just dead."
"So that kind of thing is so out of the control of the entrepreneur. And so different than what we'd like to fund. That, you know, it just scared me to death."
What does market-making look like for AI compute?
Today, GPUs are financed for years, while AI workloads often need them for just weeks. This mismatch highlights one of the biggest challenges in AI infrastructure: a fundamental lack of liquidity.
AI compute is becoming an increasingly important asset class, but the market infrastructure supporting it is still in its early stages.
At Hyperbolic, we want to provide liquidity to the global AI compute market, enabling capacity to be dynamically allocated, traded, and managed across heterogeneous infrastructure.
Excited to share our latest white paper on compute market-making, and how we think physical delivery, financial markets, and infrastructure will come together to shape the future of AI compute.
When one of the sharpest investors in AI infrastructure likes your take on compute market-making 👀
Appreciate the love, @GavinSBaker!
Jasper retweeted
We’ll be at @NVIDIAGTC Berlin! Join us for cocktails and light bites at a private indoor rooftop 🍸
We’re bringing together neoclouds, AI founders, developers, and infrastructure operators for Rooftop Cocktails & Compute. Meet fellow builders, reconnect with friends, and swap notes on open models, inference, and the compute powering AI.
🗓 October 20 · 5:30–7:30 pm Berlin time
📍 Berlin | Address shared with approved guests
RSVP here: luma.com/amnyac5s?utm_source…
#GTCBerlin
Great time at @ChronoScaleCorp’s launch party last Friday. Cool hanging out with Lawrence Lam, ChronoScale’s CTO; Sandy Gupta, VP of the Global Software Companies Ecosystem at @Microsoft; Tarkan Maner, Chair of the Board of Directors at @SVLeadershipGrp.
One idea that stuck with me: bringing AI to every company takes an ecosystem.
That resonates with what we’re building at @hyperbolic_labs. Excited to keep building alongside this community.
Managed Kubernetes is now live in beta on Hyperbolic!
We’re building a unified platform that makes fragmented, heterogeneous compute easy to access, manage, and orchestrate. This launch is another step toward that vision.
You can now launch GPU-backed Kubernetes clusters directly from the Hyperbolic console. We handle control-plane provisioning, GPU worker setup, and cluster add-ons so your team can focus on running workloads with the Kubernetes tools you already use.
Proud of the team for bringing this to life. If you’re running AI workloads on Kubernetes, give it a try and tell us what you’d like to see next!
Managed Kubernetes is now in beta on Hyperbolic.
Launch GPU clusters from the console. We handle the control plane, worker setup, and add-ons. You deploy with standard Kubernetes tools.
Explore the launch here: hyperbolic.ai/blog/managed-k…
Jasper retweeted
Great to catch up (and interview) with my friend @zjasper @hyperbolic_labs! There’s a rumor that all I do is going around taking selfies with everyone. I’d like to confirm that it’s true! 😂
Excited to be one of the launch partners! We’re building toward a future where compute is easier to access, manage, and trade.
Congrats on the launch. Looking forward to what we’ll build together!
Introducing Liquid Inference by Architect, a new LLM model router where inference providers compete dynamically to offer lowest-cost tokens. Providers onboard in minutes, not weeks. Buyers have full price visibility and receive 20% of referred fees in free inference. More info:
Liquid Inference can be used in place of any OpenAI or Anthropic API-compatible backend. Behind it, our team has leveraged our experience trading and building financial exchanges to create real-time two-sided price discovery for inference. Benefits for buyers and providers:
INFERENCE BUYERS
• Fully compatible with agentic coding tools: Claude Code, Codex, OpenCode, Cursor, Pi, Cline, and many more. Full multi-modal support.
• Hundreds of open- and closed-weight models. Create routing rule presets or use our Auto routing algorithms.
• Providers compete dynamically on price for every prompt, offering the lowest marginal cost.
• Sign up for free with email, first 500 users get $20 of free inference.
• Receive 20% of referred fees as free inference, 10% for second-level referrals.
INFERENCE PROVIDERS
• Onboard through the Liquid Inference app. We verify new providers in minutes, not weeks.
• All prompts are OpenAI API standard. Simple REST/Websocket API to register models and quotes.
• Update quotes dynamically based on your own costs. Monetize GPU capacity only when you want.
• Standardize payouts via Stripe, full itemized records of all jobs performed.
Building AI is increasingly becoming a cash flow management and risk management problem.
A few years ago, software companies could mostly treat infrastructure as a variable expense.
Swipe a credit card, spin up some cloud instances, and scale as you grow.
AI is different.
Companies are now signing multi-year GPU contracts, making large upfront payments, and committing to capacity long before they know exactly what they will need.
That creates a completely new set of questions:
• How much compute should we commit to?
• How long should we commit for?
• How much cash should we put down?
• What happens if demand changes?
• What happens if GPU prices fall?
• How do we keep committed GPUs utilized?
• What do we do with capacity we no longer need?
• How do we hedge our future compute exposure?
At some point, every serious AI company will need to do some form of compute asset management.
Not just buying compute.
Managing commitments, utilization, pricing, liquidity, and risk across the entire lifecycle of that compute.
And eventually, I think that also means hedging.
If you know you will need a lot of compute six or twelve months from now, you may want to protect yourself against prices going up.
If you already committed to more capacity than you need, you may want a way to reduce that exposure rather than simply leaving GPUs idle.
For many companies, compute may become one of the largest financial commitments on the balance sheet. Managing it well can have a meaningful impact on growth, margins, and cash flow.
At Hyperbolic, we’re striving to build both the physical infrastructure and the market infrastructure to make this work better.
Better access to compute. More flexibility around commitments. Higher utilization. Better ways to manage and hedge exposure. And ultimately, the ability to buy, use, and sell compute much more dynamically.
If you’re dealing with these questions today, feel free to reach out. Happy to compare notes.
Today, there’s a liquidity gap between compute supply and demand. Providers need upfront capital and long-term commitments. AI teams need flexibility in how much compute they buy and for how long.
We’re hiring a Capital Markets Lead at Hyperbolic to help build the financing behind the world’s compute.
At Hyperbolic, we’re turning fragmented, heterogeneous compute into one liquid AI platform. We view GPUs as valuable assets and want to provide liquidity to the people who own them and the people who need them.
You’ll work directly with me and our Head of Finance to:
• Structure debt and credit facilities that fund GPU capacity.
• Build relationships with banks, private credit funds, and institutional investors.
• Turn compute contracts into financeable assets.
The goal: make capital an accelerator for growth.
If you’ve built financing structures, raised institutional capital, and want to apply that experience to AI infrastructure, I’d love to chat.
jobs.ashbyhq.com/hyperbolic/…
We’re hiring a Quantitative Researcher at Hyperbolic.
Compute is becoming a tradable asset, but much of the market structure still needs to be built.
There is no established forward curve for GPUs. There is no liquid spot or derivatives market. Pricing varies across GPU types, regions, contract lengths, and availability.
We want to change that.
You’ll build pricing models that dynamically set spot and term rates across GPU types and regions on our platform. You’ll hedge our compute portfolio using conventional derivatives and nontraditional instruments, and design new, innovative financial products on top of our infrastructure.
If you’ve worked in quantitative research, market making, commodities, power, derivatives, or other illiquid markets and you’re excited about applying that thinking to a completely new asset class, I’d love to talk.
jobs.ashbyhq.com/hyperbolic/…
