Institutional View: Why Pump, Hyperliquid, Venice, and EtherFi Remain Undervalued

Podcast: Bankless

Compiled by: Yuliya, PANews

Recently, crypto venture capital firm RockawayX, after acquiring crypto hedge fund Relayer Capital, is raising a new $150 million liquidity opportunities fund. The fund will be led by Relayer founder and former CoinFund partner Austin Barack, focusing on uncovering undervalued crypto tokens and related equities.

In the latest episode of the Bankless podcast, Austin Barack detailed his "growth and value" investment philosophy. He pointed out that the core opportunity in the crypto market has shifted from infrastructure to applications, and highlighted why he focuses 95% of his energy on the secondary market. Through rigorous fundamental valuation models, Austin deeply analyzed the true growth potential and valuation logic of assets such as Venice, Pump, Hyperliquid, and EtherFi.

Finding Mispriced Opportunities in Crypto Through the "Growth and Value" Lens

David: Welcome, Austin. Let's talk about tokens today. First, what's your investment perspective in crypto, and what is Relayer Capital's strategy?

Austin: The crypto market changes too fast. What worked in 2017, 2021, or 2024 won't necessarily keep working. But I've found a replicable theme—the intersection of growth and value. Nobody comes to crypto looking for companies growing 10% a year at 4x earnings; you might as well buy utility stocks. Of course, with AI data center demand rising, power companies have become a bit interesting, but that's not the point.

What's truly attractive about the crypto market is that sometimes you can find projects where:

• Business is growing very fast;
• Users and revenue are expanding;
• But the market hasn't priced it highly yet.

Crypto capital flows are highly cyclical. In bull markets, many things get bid up too expensive; in bear markets, they get crushed too cheap. Because of this, the market often presents projects whose fundamentals are improving but whose prices haven't reflected it yet—that creates opportunity.

I founded Relayer about two years ago, after being a partner at CoinFund. We do both liquid and venture, but recently 95% of our focus is on liquid tokens because there are more opportunities. The two sectors I'm most focused on are crypto+AI and 24/7 tokenized trading, which covers Venice, Pump, Hyperliquid, EtherFi, and others.

David: What about the venture side? How do early-stage bets fit with the liquid strategy?

Austin: We need to distinguish here. The "growth and value" I mentioned mainly applies to liquid tokens. Because these tokens are already trading, we can see price, revenue, buybacks, burns, user growth, and other data, then judge whether the market is undervaluing them.

On the venture side, the best projects are usually not cheap; you have to pay a premium to get in. But if you can enter early (seed, pre-seed), the absolute valuation is still attractive. I still look at new brokerages, on-chain DeFi, tokenization, AI, and other sectors, but the focus is on team and execution.

David: So right now, which part of Relayer's business—liquid tokens or private venture—is drawing more of your attention?

Austin: In the first three quarters of 2024 when the fund launched, it was about 50/50. But now, I spend roughly 95% of my time on liquid tokens. The reason is that many projects in the primary market, while decent, are already growth-stage companies, more like traditional payments or fintech firms. I think they're interesting as crypto-related assets in public markets, but in the primary market, I'm not that excited. So currently I mainly focus on tokens that are already listed and trading.

David: Is this related to where we are in the cycle? After all, Bitcoin just went from $62,000 to nearly $80,000.

Austin: Not entirely. Over the past year or so, liquid tokens have been my core focus. Precisely because we went through a deep and prolonged bear market, the market has shown clear divergence. Previously, people might look at 100 tokens, but now you find that only 10, or even 5, are truly worth serious research. These projects share common traits:

• They've found product-market fit;
• Business is growing fast;
• Revenue or user metrics are clearly improving;
• But the price is still not expensive.

Of course, some assets have already rallied recently and aren't as cheap as before, but overall they remain attractive. For example, **Ethena and Pendle. I've always viewed them as "third-stage assets."** If the market bottom is stage one, and the initial asset rebound is stage two, then when on-chain yields rise and the market heats up, on-chain yield protocols like Ethena and Pendle will clearly benefit. In fact, Ethena recently rose 40% in about 30 hours, and this logic is already starting to play out.

Why Venice Might Be Undervalued?

David: You tweeted that Venice token VVV was severely undervalued at a $1 billion FDV, with a target price of $43.9. The price is higher now, but please break down this model in detail.

Austin: I come from a traditional finance background, and I built the model from scratch. Venice's business is private, uncensored AI access, aggregating various frontier and open-source models. There are two main revenue sources: subscriptions (three tiers at $18/$68/$200 per month) and additional credit purchases.

They raised equity and token financing in July at a $1 billion valuation. Venice struck a relatively elegant balance between equity and tokens. Because Venice's core business is actually off-chain. Most users just use it as a regular AI app, paying with credit cards, on computers or phones. To acquire compute, establish business relationships, and run the company, being fully on-chain would be extremely complex. So it needs an equity company entity.

But at the same time, the VVV token has its own value capture mechanisms:

• Venice uses part of its revenue to buy back and burn VVV;
• VVV also has some utility, such as related to tokenized compute;
• Long-term, the company plans to return most of its free cash flow to the token.

Currently Venice has two programmatic burn mechanisms:

• When new users subscribe, a certain amount of VVV is burned based on the subscription tier;
• When users purchase credits, a certain amount of VVV is also burned.

So when I build the model, I start with business revenue, then estimate gross margin, inference costs, marketing, customer acquisition, headcount, and other operating costs. Because Venice is not a near-100% margin business like Hyperliquid.

Assume a business with 50% gross margin, still in high-growth stage, with only 10% EBITDA margin. If it uses 8% of revenue for burns, it may actually already be returning the vast majority of free cash flow to token buybacks and burns. So you can't simply see "burn as 8% of revenue" and think that's very high value capture.

What I really focus on is: after continued investment in growth, how much cash flow remains available for burns.

• According to my model, around August 2026, Venice's annualized revenue is about $107 million, with annualized burns of about $8.3 million.
• By 2027, I project revenue to reach about $336 million, with burns reaching $70 million.

If you compare token buybacks relative to market cap to traditional PE multiples, I think a company that can grow revenue 5-10x year-over-year deserves a 50x valuation multiple, possibly even higher. $70 million times 50x gives a $3.5 billion token valuation. Then considering the token supply at the end of 2027, you get a VVV price of about $43.89. When I updated the model, VVV was around $12. Now it's around $16. I still think it's attractive.

David: What's the biggest assumption in this model?

Austin: Of the projected $70 million in burns for 2027, $29 million (about 40%) comes from their "Minds" product, which hasn't fully launched yet. That's a very significant assumption. However, I didn't make this assumption without basis.

Venice launched credit purchases earlier this year, and now its annualized revenue run rate has reached $60 million. Given the team's extremely strong product execution, projecting Minds to generate $30 million in burns by 2027 is reasonable.

David: Let me offer a counterpoint. Credit purchases, while a new feature, are essentially an extension of the existing product—they're just selling more AI usage, or letting users pay for higher usage of an existing product. "Minds" is a completely new business line (like an App Store for AI products). Credit purchases are unlikely to fail because they're essentially selling a product that already exists and has demand; but Minds is a whole new product dimension, and we don't even know if users and developers will actually embrace it.

Austin: That's a fair rebuttal. If credit purchases are a 2/10 on the new product innovation index, Minds is a 5/10.

The core of Minds is: it lets ordinary users use AI like power users. Whether it's prompt engineering, automated workflows, or code tools, Minds allows developers to build structured AI composite applications that users can use with one click.

This could also mean: I might be overestimating the burn revenue directly from Minds, but underestimating Minds' huge pull on main site subscriptions and credit consumption—because Minds makes AI more useful and easier to use, thereby boosting overall usage frequency.

David: The most exciting thing about Minds is that Venice has direct access to end users. That's the biggest difference from general model aggregators like OpenRouter. On Minds, Venice's power users can create high-quality AI composite paradigms, share them with other users, and earn revenue share. This Apple App Store-like two-sided network effect is an extremely unique bullish thesis for Venice.

Austin: Completely agree! Venice has over 4 million historical registered users, with estimated monthly active users in the millions. This highly active user base will spontaneously spread Minds apps on social media and communities because they can earn a cut. Additionally, Venice is sponsoring offline events like film festivals. In diffusion model areas like image and video generation, ordinary users really need ready-made creative toolkits like Minds.

David: Venice is a young AI startup, yet it's using revenue for token buybacks and burns instead of reinvesting all revenue into growth. That goes against startup common sense. Does that worry you?

Austin: Having a token is a double-edged sword. The upside is gaining lots of attention, fast launch, and creating new utility (e.g., users can lock VVV to mint DEM, which is essentially tokenized compute; each DEM corresponds to $1 of inference credits per day). The downside is that without clear regulatory legislation, you can't guarantee the token captures all value. Venice is very cautious: first small discretionary burns, then for new subscriptions, and now 5% burn on credit purchases. They raised $65 million, which is 10-20x the amount burned so far, so they have enough ammunition to support both growth and buybacks. This balance is very sustainable.

David: VVV's burn mechanisms are mainly new registrations and credit purchases. The team has hinted at a third possibility: renewal burns. Does your model include this?

Austin: Yes. My model is both reasonable and optimistic (about 6/10 on optimism). I expect they'll start implementing renewal burns later this year or in Q1 next year. They can start with a low percentage, observe the impact, then increase over time. I also assume the burn percentage on credit purchases will increase from the current 5% to 10% by 2027.

David: Has Venice's growth so far exceeded your expectations?

Austin: Absolutely. When I started looking at it early this year, the token was $2. At that time I estimated revenue at $10-20 million and users around 1 million. I didn't expect revenue to grow 5-10x in 8 months, users to reach 4 million, and credit growth to be so fast. Venice is one of the very few products in crypto that has truly crossed over to mainstream consumers and found PMF.

David: OpenRouter was acquired at a $7 billion valuation. What's your reaction?

Austin: This shows we're moving toward a multi-model routing world. OpenRouter leans toward the developer tooling layer, while Venice is the consumer layer. People go to Venice for privacy and to choose the best model for specific use cases. OpenRouter was valued at $1.3 billion two months ago, now $10 billion. This validates a reasonable multiple for Venice. If they continue this growth, maybe a 70x multiple is correct, which gives me more confidence in my valuation.

Fundamentals Are Decoupling Some Tokens from Macro, Exploring Three Application Growth Logics

David: Tokens like VVV and Hype have been rising on their own while Bitcoin and ETH fall. Are they really decoupled from macro?

Austin: Partially decoupled, partially coupled, but the decoupling is positive. The decoupling is because they are fast-growing businesses with fundamental value as a price floor. The coupling is because they are inherently tokens. Over the past 18 months, tokens have faced negative outflows (to US equities, AI, etc.). But I think this capital outflow is cyclical and may have reversed now. As tokens, they will benefit from more capital flowing into the crypto asset class.

For Pump and Hyperliquid, the coupling is deeper:

• Pump is strongly correlated with on-chain activity and memecoin trading. Its 90-day average revenue has grown 80% and could double or triple again.
• Hyperliquid's RWA market (commodities, stocks, indices) has high trading volume but little revenue so far; its cash cow is still the crypto token business.

If capital flows back into crypto, they will benefit in the highest-take-rate parts of their business. For example, Hyperliquid was generating nearly $1 million in daily fees a week ago, and a few days ago it generated $5 million in a single day.

Venice, on the other hand, enjoys the AI adoption wave, which is bigger than the crypto market wave.

David: Besides VVV, which other tokens excite you the most?

Austin: From a financial and valuation perspective, Pump is still very cheap. It trades at 5x buyback amount, while Hyperliquid and Lighter are around 30-40x. The market questions the durability of Pump's revenue, citing the OpenSea precedent—revenue surged then crashed 95%. But Pump's revenue has persisted for over two years and is still growing; it's not a flash in the pan. Individual meme coins may fluctuate, but Pump is "the casino for all meme coins," and that's a durable business.

Also, many people (including active crypto Twitter users) don't trade meme coins, so they struggle to understand who the users are. But casinos, lotteries, and short-dated options are all massive negative-EV industries; people participate because of variance, and there's nothing irrational about that. Pump is the crypto version of DraftKings or Las Vegas Sands.

Of course, there's still uncertainty about the equity/token split—they committed to 50% revenue buybacks for 12 months, but may not renew after that. However, for a team building a generational company, abandoning the token is not in their interest. So a reasonable buyback multiple should be 10-14x, implying Pump still has room to double.

Additionally, Hyperliquid is very interesting. It may be one of the best examples of crypto "putting the entire financial system on-chain": instant settlement, 24/7 trading, and moving all assets on-chain. New use cases are even emerging now—for example, SpaceX, Cerebras, Unitree, and other IPO companies are already seeing price discovery on Hyperliquid. I think in the future, when bankers decide an IPO price, they might directly look at Hyperliquid's trading interface: "What is the market willing to pay for it?" This could become a new price discovery mechanism.

EtherFi is another one. It was my fund's first venture investment. They've pivoted from staking to yield products, credit cards, and now a mature neo-brokerage. The team's execution is extremely strong; you can trade any on-chain asset and borrow/lend. It was originally valued by the market as a liquid restaking business like Lido, peaking at $8 billion FDV, then fell as restaking hype cooled. But today EtherFi's business is fundamentally different—65%+ of revenue comes from the new banking/brokerage business (credit card fees, lending interest), and only 35% from staking yield. And this ratio is still shifting, with the new brokerage part growing faster.

On valuation, it's currently around 10-15x earnings. For a business growing 10x (credit card daily volume from $300k to $3-4 million) and just starting programmatic buybacks, with the token nearly fully circulating and no emission pressure, I think it's undervalued. Blockworks did a model assuming growth halves, getting $21 million in 12-month buybacks; I assume $30 million, give it 30x, which corresponds to a token price over $1 (double the current price), not even counting multiple expansion as a sector leader.

David: EtherFi fits the modern startup model: small team, leveraging big technology. Because of tokenized assets, they evolved from Neo Bank to Neo Brokerage at almost no cost—Ethereum does a lot of the legwork for them.

Austin: Right. It currently has only $20 million in lending, with interest income at just 4%. They leverage existing DeFi infrastructure, like partnering with Aave to run their own Aave v4 instance with an 80/20 revenue split (80% to EtherFi). Traditional neobanks (like NuBank) have interest income at 60-70%. From that perspective, there's huge room to grow.

The Next Wave of Crypto Opportunities Belongs to Applications

David: Looking ahead to 2026-2027, how will this cycle be defined?

Austin: We've completely left the "super infrastructure era" behind. Excitement about new L1s is a thing of the past. An interesting data point: in the past, execution-layer infrastructure accounted for over 95% of crypto revenue; now applications account for 2/3 and execution layer 1/3. I believe over 90% of revenue will be generated by applications in the future, and the most durable tokens will be "applications" and "money."

Bitcoin won't disappear, but OG privacy coins like Zcash are attracting long-time Bitcoin holders, returning to crypto's original ethos and drawing capital inflows.

Ethereum's potential as "money" has also rekindled my interest (considering Bitcoin's quantum risk and concentrated ownership risk).

At the application layer, Solana is the most active blockchain, and it enabled Pump. While it no longer has the MEV revenue it once did, it remains one of the strongest bets on crypto adoption.

Ultimately, the best buys in 2026 will be concentrated in those 0-to-1 applications—ones that have found the intersection of crypto and the real world, as well as assets truly recognized as "money." Venice, Hyperliquid, Pump, EtherFi, Bitcoin, Zcash, and others will all look like incredible opportunities in hindsight.

Author: Yuliya

This article represents the views of a PANews columnist and does not reflect PANews' position, nor does it assume legal responsibility.

The article and opinions do not constitute investment advice.

Image source: Yuliya. If there is any infringement, please contact the author for removal.