Dear Partners,
First, a few administrative updates. We will be sending 2025 audited financials by the end of April. Capital account statements and K-1s have already been distributed. Please reach out with any questions.
EV3 was incorporated almost four years ago. Our mission has not strayed: to become the world's leading early-stage investor in crypto-enabled businesses and deliver exceptional long-term returns for ourselves and our investors. To date we have raised $100M from a set of world-class partners and invested $35M into early-stage crypto projects. We specialize in partnering with entrepreneurs who harness the power of network economies to disrupt legacy $1T+ industries like advertising, computing, energy, and telecom.
The past few months have kept us extremely busy. Since our last investor letter:
- We made two new investments from Fund II, both arising from our onchain credit thesis.
- We announced our $62M Fund II with an exclusive article in Fortune and published our annual State of DePIN report with Messari, reaching 750K+ social impressions across our releases.
- We taught eight classes at Harvard Business School, bringing founders and executives from Hyperliquid, Helium, Worldcoin, and others to inspire the next generation of crypto entrepreneurs.
- We launched the EV3 Fellowship and interviewed hundreds of exceptional young students exploring the edges of rocket design, synthetic cells, microfluidics, and everything in between.
Zooming out, today's world is increasingly characterized by accelerating rates of change. Since our last letter, regime change in Venezuela and a war in the Middle East has disrupted global commodities flows. The unrelenting buildout of AI infrastructure has driven massive deflation in the cost of intelligence whose impact is beginning to be felt on labor and software markets. As if that isn't enough tail risk materializing, prediction markets are beginning to price in the existence - or at least the confirmation - of alien life.
In chaotic (& chaordic) eras, the firms that thrive aren't the biggest, or even the smartest, but the ones that adapt most effectively to rapidly-changing environments. Today's letter covers our views on the state of broader crypto markets, as well as a review of the Fund I portfolio and Fund II new investments.
As always, we welcome feedback and questions on our thinking, especially from our trusted partners.
Your partners,
Sal and Mahesh
Public Markets
Crypto ended the quarter with a $2.4T market cap, down -21% QoQ and -39% since our prior letter, underperforming both equities and commodities.1 Zooming out, this is roughly the halfway point between the peak ($4.3T in Q4'25) and trough ($0.8T in Q4'22) of crypto market cap since the launch of EV3.
In our Q2'25 letter, we discussed the three sources of capital inflows into crypto: TradFi institutions, retail, and protocols. TradFi net sold -$20B worth of crypto majors in the past six months after net buying +$90B over the preceding six. Retail sentiment is showing some signs of reversing (e.g., Robinhood users net bought roughly +$3.6B of crypto in Q4'25, after net selling -$1.4B earlier in the year), but it was not nearly enough to offset the onslaught of institutional sellers. Falling valuations beget declining trading volumes, which caused the buyers of last resort - the protocols themselves - to pull back, further exacerbating the downturn. The two most profitable crypto apps, Hyperliquid and Pump, reduced buybacks by $150M over the past six months vs the prior six, after earnings, driven by trading volumes, declined by almost -20%.2
We believe these dynamics are close to bottoming if they haven't already. When it comes to institutional selling, the worst is behind us: the two largest DATs, Microstrategy and Bitmine, hold >80% of DAT assets and are valued near par (0.92-0.94x) with continued access to credit markets. Smaller DATs will struggle, but their overall impact on flows will be limited. Between Mahesh's Director role at Strive, a leading Bitcoin treasury company, and our TradFi networks we have had a front row seat to the scale of demand for ~10% yields from perpetual/preferred structures issued by large DATs. Several DATs are routinely placing nine-figure debt issuances on monthly cadences. Unlike BDCs and other private credit instruments facing redemption pressures, the large DATs are - for better or worse - structured in a way where they almost never have to redeem underlying crypto exposure (we think a much more likely outcome is DATs selling crypto to diversify into cash-flowing assets like real estate and commodities). Looking back on the "DAT experiment" of the past eighteen months, it's obvious that the bubble and subsequent washout led to real financial engineering: TradFi hedge funds, via structured products, now have a strong and persistent bid for crypto, albeit only for the top ~3 assets. We expect flows from DATs, ETFs, and 401Ks to flip from negative to positive this year thereby removing the biggest structural headwinds to crypto prices.
There are signs that protocol buybacks are bottoming too. Beyond DePIN, crypto revenues are largely driven by trading activity and are therefore highly reflexive to token prices. That said, Hyperliquid and Pump earnings are only down -19% since our prior letter, while crypto market cap is down -39%. This is a result of these projects shipping new, non-crypto indexed products like commodities and equities trading. These names trade at 14x and 2x earnings (buyback) multiples today, respectively, as investors apply a discount for cyclicality of crypto trading volumes. However, their revenue composition is quickly changing: non-crypto products grew from zero to 40% of Hyperliquid's daily trading volume since our last letter. Over the coming year, we believe the market's perception towards the highest-grossing protocols will shift from "cyclical apps" to diversified, high-growth trading engines. Re-rating to higher earnings multiples will help capitalize these protocols for the next leg of growth and kickstart a new wave of reflexivity.
These dynamics are forcing crypto investors to choose between two diametrically- and philosophically-opposed poles. On one hand, you can bet on the TradFi flows reversing for the most liquid crypto-assets: $BTC, $ETH, $SOL. The structured notes underlying DATs have proven product-market fit with volatility-harvesting hedge funds, and the most egregious excesses (by book value premiums) have washed out and today represent a small and shrinking share of DAT capital. The downside is that betting on L1s requires paying several hundred times declining earnings at a time when stablecoins, dapps, exchanges, and validators are all looking to internalize a piece of their margin pool. Very simply, there is no DCF-based argument you can make for L1s at today's prices: they are either a 3-6 month bet on TradFi flows reversing, or a 3-6 decade bet on a monetary premium emerging. Nothing in between.
On the other hand, you can bet on the app-layer with protocols whose revenues are growing rapidly and are increasingly non-cyclical. These protocols have a track record of capital return in the form of token buybacks and trade at single-digit to mid-teens multiples—a meaningful discount to web2 comps. On a 2-3 year time frame, you'd be hard-pressed to find any asset, anywhere in the world, trading at lower growth-adjusted cash flow multiples than the leading crypto alts tokens. But there's a drawback: there's no one to take you out of the trade! Institutions only touch the most liquid of tokens, i.e. L1s. Crypto-native funds have been decimated by LPs pulling back and are largely absent from the liquid alts market. Non-crypto funds are too spooked by value accrual conflicts (tokens vs equity) to play ball. Retail investors will largely buy whatever Robinhood puts on their homepage, which today is prediction markets. Put simply, there are no discriminatory (fundamentals-based) buyers left for quality crypto-assets below $10B market cap. The only buyers you can count on are flows-based. If you've had the (mis)fortune of being a value investor anytime in the past decade, you'll know the feeling.
Given the playing field, liquid crypto investors can be loosely bucketed into one of three categories:
- Those betting on L1 tokens with a 3-6 month horizon, with no clue what the assets are worth.
- Those betting on dapps with a 3-6 year horizon, with no clue as to how long they'll have to wait.
- Those betting on crypto equities, getting adversely-selected against crypto-native protocols.
In other words, investing in liquid crypto today requires taking a flyer on value, duration, or adverse-selection. We are firmly in category #2, betting on protocols that are clear category leaders with strong forward growth prospects and proven capital return mechanisms, even if it's not obvious when meaningful liquidity will emerge. There are reasons to believe it may come sooner rather than later, including the recent SEC & CFTC joint guidance which provides a pathway to resolving token vs equity conflicts that keep both crypto-native and non-crypto-native investors from investing in alts today.
For investors that are willing to take on the duration risk, today's alts market is ripe with potential value. Pump and Layer3 are trading at or below cash, while generating significant earnings.3 Hyperliquid trades at half the earnings multiple of Robinhood and generates 100x higher earnings-per-employee. Helium and Geodnet trade at ~10x earnings after having grown between 2.5x and 7x over the past year. Post-2022 DePIN launches now almost all trading below their seed valuations, and a handful are finding product-market fit. These tokens are largely past the insider unlock overhangs that have plagued alts in the past.
Below is a brief update on the networks we are most excited about.
Helium ($HNT)
Helium ($HNT) is down -55% since our last letter, currently trading at a $200M market cap and 12x burn multiple. Fundamental network activity was up +150% over the same time period, with daily active users reaching 3.5M and daily traffic reaching 120 TB. AT&T and T-Mobile pay $17M per year to offload their customers' cell phone traffic onto 43K community-owned WiFi hotspots. The network is net deflationary, meaning they no longer subsidize miner earnings with inflationary token incentives. By almost any account except token price, the network is firing on all cylinders and the DeWi thesis is playing out.
Helium has become one of the posterboys for the equity vs token conflicts plaguing most tokens (ex L1s). Nova Labs, the development company that created Helium, raised money from both web2 and web3 VCs. Some of these VCs invested exclusively in Nova Labs shares, some exclusively in $HNT tokens, and many in both. This created a situation that persists today whereby different investors have a fiduciary duty to drive value to equity vs tokens. This is problematic because 1) Helium Mobile, the subscriber business owned by Nova Labs, is a customer of the network, and 2) in an M&A scenario, it's unclear how proceeds would be split. Given founders' primary legal fiduciary obligation is to shareholders, many otherwise-eager token investors have stayed on the sidelines. In our view, the most likely outcome is a merger of the cap tables where Nova Labs shareholders are bought out with newly-minted $HNT. However this will require the market to digest a one-time inflation shock, and it's unclear on what timeline a merger could be approved or executed. Assuming a simple 50/50 split, $HNT's pro forma multiple on run-rate earnings would be 25x today (excluding the subscriber business entirely), with +600% LTM earnings growth.
Grass ($GRASS)
Grass ($GRASS) is down -61% since our last letter, currently trading at a $160M market cap and $280M FDV. This represents a multiple of 3x and 5x, respectively, on Q4'25 annualized (offchain) revenues. In December, we sat down with the Grass team under NDA, reviewed their customer contracts, and verified $13M of billed revenues in Q4'25 (>$50M annualized) from leading AI labs. Grass's 3M users run nodes that scrape data from the public web and allow AI labs to avoid the operational complexity and also liability of scraping data themselves. While we cannot disclose Grass's specific customers, they include several of the labs you have heard of whose models you might be using in another tab right now.
Like Helium, Grass suffers from investor perceptions of equity vs token conflicts. They do not publicly disclose customer names or demand-side metrics, nor conduct large-scale token buybacks. Despite that, we think these concerns are misplaced. The former (lack of disclosure) is due to customer concentration, i.e. Grass would weaken its negotiating leverage with the largest AI labs by disclosing its aggregate scale. The latter (lack of buybacks) is due to Grass re-investing operating cash flows into building bare metal cloud infrastructure to improve the margins on its current products and enable new products like LCR. Given we've seen the contracts with our own eyes, we are inclined to believe the revenues are real. If true, a category-leading token with direct exposure to leading AI labs is trading at 3-5x revenue multiples.
Geodnet ($GEOD)
Geodnet ($GEOD) is down -9% since our last letter, currently trading at a $60M market cap and 8x burn multiple. Fundamental activity was up +50% over the same time period, with customers paying $7M per year to access GPS signals from satellites captured by Geodnet's 20K GNSS ground stations. Customers are agriculture, construction, and robotics companies with outdoor machines that use data from nearby Geodnet stations to improve GPS accuracy by 100x, from meters to centimeters. The company has some exciting initiatives slated for 2026, including the launch of a low-cost US manufactured drone.
Geodnet is the steady compounder of DePIN. They've grown revenues every single quarter since launch, compounding at over 40% QoQ. Like Helium, $GEOD is nearly fully-distributed and will likely become deflationary this year. $GEOD's issue is liquidity: daily trading volume is <$250K which prevents trading in any meaningful size. While it's not clear to us how or when the liquidity issue gets fixed, today's market cap represents a <1x multiple of Q4'2027E token burn if Geodnet continues growing at its historical rate.
OverTheReality ($OVR)
OverTheReality ($OVR) is down -69% since our last letter, currently trading at a $4M FDV and a 10x multiple of (offchain) revenues. OVR's business is similar to Grass, except that instead of scraping digital data from the web, OVR sells photos and videos that its users collect via smartphone cameras. This data is particularly useful for the emerging crop of labs focused on next-generation world models. The network launched in 2020 as an NFT-based AR/XR platform and "accidentally" built one of the largest real-world image datasets in the world: 95M images, across 200K locations, representing 900 TB of data.
OVR is the sleeper pick of DePIN. The team is based in Italy and has largely stayed under the radar of crypto investors, but they are beginning to sell data to the same set of customers as Grass while trading at 1% of Grass's market cap. The 10x run-rate multiple is based on early pilots with large labs that can expand meaningfully, plus data can be resold to several labs. Grass grew revenues from $2.8M to $12.8M from Q2 to Q4 2025 while serving a similar set of customers, with similar data, on similar terms.
MetaDAO ($META)
MetaDAO ($META) is up +25% since our last letter but down -75% from its all-time-high in January, currently trading at a $75M market cap and 25x revenue multiple. MetaDAO is a capital-raising platform or launchpad that uses a mechanism called futarchy to programmatically enforce tokenholder rights, like a board of directors. Tokens launched on MetaDAO trade on a native DEX, which earns $3M run-rate fees for $META tokenholders. Since our last letter, 10 startups have collectively raised $30M on MetaDAO.
The main problem with new launchpads is adverse selection. MetaDAO is beginning to show signs that its architecture and brand are attracting strong, ideologically-aligned founders and repelling all the rest. In February, tokenholders voted to liquidate the treasury of Ranger Finance, a project that had raised $8M under allegedly false pretenses (overstated revenues)—a clear disincentive against teams exaggerating their traction while raising capital on MetaDAO in the future. At the same time, the quality of teams launching on MetaDAO is rising, i.e. Multicoin-backed P2P launched last week. Leading launchpads like Pump and Virtuals went through bursts of extreme reflexive growth. We think $META will do the same.
Bittensor ($TAO)
Bittensor ($TAO) is up +1% since our last letter, currently trading at a $2.9B market cap and $6.4B FDV. Bittensor is a network-of-networks, aggregating various domain-specific AI "subnets" behind a single unified token economy. The leading subnet, Chutes, generates $5M in annualized revenues serving AI inference workloads. There are 125+ other subnets across various domains and levels of traction.
Our $TAO thesis has three legs. First, we believe that Bittensor unlocks a new class of entrepreneurs - mechanism designers - that are brilliant at designing incentives but lack the capabilities to raise capital, acquire customers, and operate infrastructure in a traditional startup. Second, we believe these new entrepreneurs, armed with open source AI models, will create products competitive with closed source AI labs. Lastly, we believe that as thesis this plays out, $TAO's Bitcoin-like philosophy and governance will drive markets to re-rate it to valuations comparable with leading AI labs. With $250-500M in daily trading volumes, $TAO remains the single most liquid token across all of AI and DePIN.
Notes
As a reminder, we publish our State of DePIN report in lieu of a Q4 letter each year. ↩
TradFi institutions inferred from publicly-reported ETF and DAT flows. The two most profitable networks are Hyperliquid and Pump. ↩
Refers to token market cap as a multiple of cash at the development company's balance sheet (not protocol-owned value). ↩