Yan Liberman
Key Views & Dialogues
The Secret Supplier Feeding Frontier AI
Grass uses consenting residential bandwidth and browser infrastructure to reach sites blocking data-center traffic; Andrej Radonjic says it has delivered exabytes to AI labs. Labs value targeted signal and speed: one discovery found 0.16% useful, while sophisticated buyers link data value to benchmark gains, token use and returns. Content API arrived ahead of the previously discussed 2027 timetable, but regional capacity and latency remain inference-era constraints.
View summary (7)
- Grass’s advantage starts with getting AI systems onto websites that block data-center traffic. Andrej Radonjic says routing through residential IPs is the best way to get broad access, with browser infrastructure handling additional defenses. Grass pays consenting participants for bandwidth; Andrej says it has delivered exabytes of data to AI labs.
- Frontier labs increasingly buy useful signal and speed alongside raw volume. A lab’s discovery process can identify a small share of a dataset—for example, 0.16%—as useful, driving more targeted purchases. With new models resetting benchmarks every few months, customers risk being “left behind”; sophisticated labs estimate data’s value through benchmark improvements, subsequent token usage and financial returns.
- Robotics creates demand for carefully selected behavioral video. Andrej updates Yan’s roughly 1 billion figure to more than 100 billion indexed videos. Yan points out that indexing does not mean downloading every video. Generative video benefits from diverse examples that can be combined, while robotics currently needs footage that teaches actions and interactions—a harder filtering problem than simply accumulating video.
- Training crawls are building the infrastructure for live inference and search. Andrej describes training-data collection as a “treasure hunt” that also maps the web. Grass has released its Content API under Live Context Retrieval; Search is planned next, while he expects training and inference data eventually to converge.
- Inference makes regional capacity and latency increasingly important. Andrej previously thought Live Context Retrieval needed ten times as many nodes, but infrastructure improvements enabled an earlier launch than the previously discussed 2027 timetable. More residential nodes are expected as regional demand grows, and the “speed of light” constrains how close routed requests can get to original latency.
- Grass’s customer business and intellectual property sit within an ownerless foundation. “Nobody owns it,” Andrej says; Wynd Labs supplies services, employs staff and holds no IP. The revenue attestation reconciles foundation purchase orders with cash receipts, including payment delays such as 60 days; it concerns the foundation rather than a separate equity vehicle.
- Yan recounts a founder’s view that superior reinvestment returns can constitute a moat, and Andrej agrees. Yan cites the AI-data market estimate of $100 billion by 2030; Andrej sees expansion opportunities in synthetic traces, reinforcement learning and other data categories. The closing moat discussion centers on getting more out of each dollar than competitors do.
The Six Tokens To Own This Cycle | Yan Liberman
Yan sees Bitcoin reaching the high $90,000s before materially easier liquidity, with only 10–15 investable tokens. His six-token book favors uncrowded productive assets: Grass has roughly $70 million expected revenue and $40 million indicated operating profit against about $330 million FDV. Grass’s October unlocks and AERO’s second-half-of-September expansion are catalysts, but demand, emissions and cross-asset opportunity cost remain unresolved.
View summary (7)
Yan Liberman is long the rally but does not see near-term Bitcoin highs without materially easier liquidity. The squeeze began from “bottomy” positioning plus the prospect of long-end curve control resembling QE; subdued leverage and Saylor’s absence are constructive, but Yan sees Bitcoin reaching the high $90,000s before needing a more accommodative Fed or favorable inflation data. “It kind of has to be in this order”: slow Bitcoin grind, pause, then alts.
The opportunity set has narrowed into a stock-picker’s market of perhaps 10–15 investable tokens. Obvious revenue names are already crowded and need growth to expand their multiples, so Yan wants assets that “look and smell like those” without the positioning. HYPE remains a comfortable long-term hold because of non-crypto trading activity and a prospective USDC bid supporting buybacks, although $200 within four months “would be really surprising.”
His six-token portfolio comprises HYPE, Zcash, Lighter, Venice, Grass and AERO, likely Aerodrome, with sizing undisclosed and positions split between spot and leverage. The preference is for productive revenue, especially demand not wholly dependent on crypto activity; Yan typically adds leverage early for “torque,” then unwinds it into spot after a substantial move. He distinguishes durable earnings from assets whose fundamentals are tightly linked to market reflexivity.
Grass is Yan’s clearest forgotten-business repricing thesis: roughly $70 million of expected revenue and $40 million of indicated operating profit against about $330 million FDV. Its roughly eight-million-node residential-IP network supplies continuously refreshed web data to frontier labs, while a planned Live Context Retrieval subscription could unlock real-time sites blocked to ordinary agents. Investor unlocks are scheduled to be done in October, roughly 30% of supply sits with the foundation, and there is no parallel equity claim: the token is the value-accrual mechanism.
AERO is an event-driven trade with a dated catalyst and explicit invalidation. Its delayed expansion is slated for the second half of September, with Ethereum and Arbitrum announced and Robinhood unconfirmed; Yan expects broader distribution to lift TVL, volume and fees quickly. Planned dynamic emissions and new revenue lines—estimated to add 40% without an emissions offset—could answer the longstanding criticism that token emissions neutralize staking revenue.
Delphi finds more opportunities in liquid crypto than new crypto venture, while its sole recent crypto startup investment targets a dollar-hedged Turkish-lira carry trade yielding about 10.5%. Tory Finance offers organic yield without token incentives and potentially large capacity. The conversation flags the dollar hedge and bank-counterparty structure, while users could add another carry layer by borrowing USDC against the position. Only one of Delphi’s five current venture investments is crypto, reflecting a broader-adoption thesis that has “fallen by the wayside” as AI absorbs talent, attention and capital.
The hardest pushback is opportunity cost: productive tokens remain among the highest-beta assets and must compete with Nvidia, SpaceX and other scalable businesses. Yan wants cash available for a brutal economic unwinding and asks what tangible flows and adoption would justify HYPE at $79 clearing a 2–3x return. The hosts’ corrective is blunt: “Do you want to be right? Do you want to make money?” Both AI private markets and crypto tokens can be simultaneously frothy.