Multicoin invests on the order of $1.6 billion, and what makes them worth studying is less any single position than the method: where an investor's edge actually comes from, how you value things nobody knows how to value, and why the best entry points rarely feel comfortable.
1. Know exactly where your edge is
The first question any investor should be able to answer is where they have asymmetric edge. For Multicoin partner Shayon Sengupta, the answer is depth rather than stock-picking. His real advantage is the sheer amount of time he spends inside his companies, working on product and distribution, sitting in the hard board meetings, doing incubations and turnarounds. Multicoin's structure is what lets that happen: from day one, everyone is handed as much agency as they will take, and is expected to define their own way of producing returns. Picking is table stakes. Knowing what you are uniquely good at, and building your whole process around it, is the edge.
2. The money is made in the bear market
There is a line from an old rowing coach worth keeping in mind: races in the spring are won in the fall. The point is that the work that produces returns happens long before the returns show up, and specifically in the quiet, unglamorous part of the cycle.
All the money actually gets made in the bear market. Shayon Sengupta, Multicoin Capital
The easy money has been competed away. The old pattern of raising at one price, then a higher price, then a listing, and dumping into it, is over. What replaces it is longer time horizons and stranger deal structures: buyouts, incubations, contributor arrangements. The conditions for real early-stage work are better now precisely because everyone else is asleep at the wheel.
3. Crypto is money, so use it to coordinate real work
Once you strip away the ideology, the revealed preference is clear: people treat these systems as money. The interesting question is what you do with an asset ledger beyond payments and trading. The answer Multicoin lands on is incentive alignment, using tokens to pay people to solve problems that were previously uncoordinated. That is the through-line from its DePIN thesis to what it now calls internet labor markets: give people tokens to do some arbitrary useful thing, and you can bootstrap a network that could not otherwise exist.
The canonical early example was Render Network, one of the first positions he worked on, back in 2021. The insight was that compute is not homogeneous, so the winner would be whoever could aggregate the right resources and orchestrate the workloads across them. Render and its parent OTOY, led by Jules Urbach, were vertically integrated in a way no other compute network was. The thesis has since matured from static, one-time contributions, like plugging in a hotspot, into ongoing active-contribution networks such as Fuse and Crunch.
4. Value networks like businesses, then force the market to agree
Valuation is where it gets sharp. DePIN tokens trade badly, for two fixable reasons:
- Emissions. Time-based schedules dump a fixed number of tokens on contributors every epoch, most of which get market-sold, and no chart survives that forever. The better teams are moving to dynamic rewards tied to the actual quality of work.
- Valuation. The market does not know how to value these networks, and that is partly the teams' own fault.
The fix for the valuation problem is web2 comparables. Is Helium priced like Crown Castle or like T-Mobile, or some blend? Is GEODNET a Trimble, a Hexagon, or a TomTom? A team cannot just complain that the market misunderstands it. It has to say, explicitly, here are my three metrics, here is how to think about demand two years out, here is how you should price me. That investor-relations muscle barely exists in crypto, and building it is an edge.
GEODNET is the sharpest case study. Multicoin passed in 2022 on market-size and execution concerns, then re-engaged in 2024 when token burns went vertical and led an $8M round into founder Mike Horton's network. What changed his mind was watching Horton do the hard physical thing: take big, expensive positioning hardware and make it small and cheap, cover most of the addressable world in under eighteen months on a single-digit share of token supply, and pull enterprise customers off the incumbents at an order-of-magnitude lower price. With millions of robots, drones, and autonomous vehicles all needing centimeter-level positioning, that footprint is the bet.
5. Back mission-critical infrastructure, run by teams you would not bet against
On Jito, the thesis is that it provides a mission-critical service to Solana: getting transactions reliably included and ordered. As real economic value on Solana grows, Jito compounds with it, and the DAO and foundation are direct beneficiaries. The honest part is the risk. The protocol is changing, Jito's newer work like BAM (its Block Assembly Marketplace) is partly a bridge, and the block engine's role could shrink as Solana internalizes more of these functions. The conviction rests less on the current mechanism and more on the team behind it, led by Lucas Bruder.
I don't know why you would bet against that team. Shayon Sengupta, on Jito
6. Be willing to say what is overrated
The other side of the ledger, what looks overrated:
- Meme coins. They have stagnated, and assets with a three-hour shelf life are not interesting for global finance. He would rather see the engineering talent that went into making meme coins tradable pointed at longer-lived, higher-quality assets, the real-world risk of FX, commodities, and equities coming on-chain and trading through every conceivable front end.
- Stablecoin layer-1s. If all you do is payments there is little economic value to capture, and liquidity network effects favor the incumbents, Tether and Circle.
The throughline across all of it is that Multicoin treats investing as a craft rather than a series of calls. The edge is operational. The discipline is valuing what you build like a real business and telling the market how to price it. And the best time to do the work is exactly when it feels least rewarding.