Crypto venture firm RockawayX is raising a $150 million liquidity opportunities fund after acquiring crypto hedge fund Relayer Capital. The new vehicle is led by Relayer founder and former CoinFund partner Austin Barack, with a mandate focused on undervalued crypto tokens and related equities.
Speaking on a recent episode of the Bankless podcast, Barack laid out what he called a "growth and value" approach to crypto investing. His central point was simple: the most attractive assets in crypto are not slow-growing, low-multiple businesses, but projects where users, revenue and operating momentum are expanding quickly while market pricing still lags the underlying fundamentals.
Barack said Relayer was founded about two years ago. The firm invests across liquid markets and venture, but today about 95% of his attention is on liquid tokens. He said the opportunity set is strongest there right now, particularly in two areas: crypto plus AI, and tokenized 24/7 trading. Venice, Pump, Hyperliquid and EtherFi were the main examples he used.
Why liquid tokens now dominate his focus
Barack said each cycle rewards a different playbook. What worked in 2017, 2021 or 2024 does not automatically keep working. What has stayed useful, in his view, is the overlap between growth and value.
Crypto, he said, is highly cyclical in the way capital moves. In bull markets, many assets become too expensive. In bear markets, some get pushed too low. That creates situations where fundamentals improve while price has not caught up yet.
He drew a clear line between venture investing and liquid token investing. In liquid tokens, investors can track market price, revenue, buybacks, burns and user growth in real time, then judge whether the market is mispricing the asset. In venture, he said, the best companies are usually not cheap, and the main way to get attractive absolute valuations is to enter early at the seed or pre-seed stage.
During the first three quarters of 2024, Barack said his fund's time was split roughly 50-50 between private venture and liquid markets. That has now shifted dramatically. About 95% of his time goes to liquid tokens. He said this is not just a reaction to Bitcoin moving from $62,000 to nearly $80,000. Over roughly the past year, liquid tokens have been his core focus because the long and deep bear market created meaningful dispersion. Instead of a universe where 100 tokens all look worth studying, he said the list may now be closer to 10 names, or even five.
Those names, he said, share several traits: they have found product-market fit, they are growing fast, their revenue or user metrics are improving, and they are still not expensive. He added that some assets have already moved off the lows and are no longer as cheap as before, but still look attractive on a broader view. He cited Ethena and Pendle as examples of what he called "phase three assets." In his framework, the market bottom is phase one and the first rebound is phase two. Once onchain yields rise and markets heat up, yield protocols like Ethena and Pendle benefit more directly. He noted that Ethena had risen about 40% in roughly 30 hours, which he saw as the thesis starting to show up in price.
Venice: subscriptions, token burns and a $43.89 model
Among the names discussed, Venice and its token VVV received one of the most detailed breakdowns. Barack said he had posted earlier that VVV looked deeply undervalued at a $1 billion fully diluted valuation, with a target price of $43.9. When he updated that model, VVV was around $12. By the time of the podcast, it was around $16, and he said he still viewed it as attractive.
Venice, he said, offers private, censorship-resistant access to AI by aggregating frontier and open-source models. Revenue comes mainly from two sources: subscriptions and additional credit purchases. The subscription tiers are priced at $18, $68 and $200 per month.
He also said Venice raised equity and token financing in July at a $1 billion valuation. In his view, the company has struck a relatively elegant balance between the offchain business and the token. Most users simply treat Venice as a normal AI app, pay with a credit card, and use it on a computer or phone. Running the business, securing compute and building commercial relationships all fit better inside a conventional company structure. At the same time, VVV has its own value capture path.
- Venice uses part of its revenue to buy back and burn VVV.
- VVV has utility tied in part to tokenized compute.
- Over the long run, the company plans to return most free cash flow to the token.
Barack said Venice currently has two programmatic burn mechanisms. New user subscriptions burn a certain amount of VVV depending on the plan selected. Credit purchases also burn a certain amount of VVV.
He said his model starts from revenue, then works through gross margin, inference costs, marketing, customer acquisition and labor. Venice is not a business with economics close to Hyperliquid's, where margins approach 100%, so he argued that investors should not look at a burn rate as a share of revenue and immediately conclude that value capture is unusually strong.
His example was that if a company has a 50% gross margin, remains in high-growth mode, and only generates a 10% EBITDA margin, then allocating 8% of revenue to token burns could already mean it is returning most of its free cash flow to buybacks and burns. The key question, he said, is how much cash remains for burns after the business continues investing in growth.
Under his model, Venice reaches annualized revenue of about $107 million around August 2026, with annualized burns of about $8.3 million. By 2027, he projects revenue of roughly $336 million and annualized burns of about $70 million. He then maps buybacks relative to market value to something like an earnings multiple. For a business growing revenue 5x to 10x year over year, he said a 50x multiple is reasonable, and possibly low. Applying 50x to $70 million implies a token valuation of about $3.5 billion. Using end-2027 token supply, he arrives at a VVV price of roughly $43.89.
Barack also identified the biggest assumption in that model. Of the projected $70 million in burns for 2027, $29 million, or roughly 40%, comes from the not-yet-fully-launched Minds product. He acknowledged that this is a major assumption. Still, he said it is not baseless. Venice introduced credit purchases earlier this year, and the business is now running at an annualized revenue rate of $60 million. Given what he described as strong product execution, he said it is reasonable to expect Minds to generate about $30 million of burns in 2027.
Bankless host David pushed back on that point. Credit purchases, he said, are a new feature but still an extension of the existing product. Minds is different. He described it as a new line of business, closer to an app store for AI products, where it is not yet clear whether users and developers will respond in the same way.
Barack said the criticism was fair. If credit purchases are a 2 out of 10 on the scale of product novelty, he said Minds is more like a 5 out of 10. In his description, Minds lets developers build structured AI bundles that package prompt engineering, automated workflows and coding tools into one-click experiences for ordinary users. That means, in his view, he may be overestimating the direct burn revenue generated by Minds while underestimating the boost Minds could provide to core subscriptions and credit consumption by making AI easier to use.
David added that the more interesting part of Minds is Venice's direct reach to end users, which separates it from general model aggregators such as OpenRouter. On Minds, Venice power users can create high-quality AI bundles, share them with other users and earn revenue splits. He compared that to a two-sided network effect similar to Apple's App Store.
Barack agreed and said Venice has more than 4 million historical registered users, with monthly active users estimated at more than 1 million. Those engaged users, he said, have a built-in incentive to promote Minds applications across social media and communities because they can earn from distribution. He also noted Venice is sponsoring offline events including film festivals, and argued that in areas such as image and video generation, ordinary users have a clear need for ready-made creative toolkits.
David raised another issue: Venice is a young AI startup, yet it is using revenue for token buybacks and burns instead of reinvesting everything into growth. That runs against standard startup logic.
Barack answered that tokens are a double-edged sword. On the positive side, they attract attention, accelerate go-to-market and create new utility. He gave one example: users can lock VVV to mint DEM, which he described as tokenized compute, with each DEM corresponding to $1 per day of inference credit. The downside is that without clear legislation, there is no guarantee that tokens will capture all the value. He said Venice has taken a cautious approach so far, starting with small discretionary burns, then adding burns tied to new subscriptions and now a 5% burn on credit purchases. The company has raised $65 million, which he said is 10x to 20x the amount already burned, giving it enough balance sheet capacity to support growth and buybacks at the same time.
On subscription renewals, Barack said his model already assumes a third burn mechanism will be introduced. He expects Venice could start renewal burns later this year or in the first quarter of next year, beginning at a low rate and scaling it over time. He also assumes the burn rate on credit purchases rises from 5% now to 10% by 2027.
He said Venice has already exceeded his expectations. When he started paying attention earlier this year, the token was around $2, revenue looked to be around $10 million to $20 million, and the user base was about 1 million. Instead, revenue grew 5x to 10x within eight months, users reached 4 million, and credit growth accelerated as well. He described Venice as one of the few products in crypto that has truly crossed into mainstream consumer usage while also finding product-market fit.
He also commented on OpenRouter being acquired at a $7 billion valuation. To him, that supports the idea that the market is moving toward a multi-model routing world. OpenRouter is more of a developer-tool layer, while Venice is consumer-facing. He said OpenRouter was valued at about $1.3 billion two months earlier and now stands at $10 billion, which in his view supports the multiples he is applying to Venice. If Venice keeps growing at this pace, he said, even a 70x multiple may not be out of the question.
Partly decoupled from macro, but still tokens
Asked whether names such as VVV and HYPE have truly decoupled from macro because they can rise even when Bitcoin and ETH fall, Barack said the answer is both yes and no.
The decoupling, he said, comes from the fact that these are fast-growing businesses with fundamental value that can provide a floor. The coupling remains because they are still tokens, and tokens spent the past 18 months fighting capital outflows toward U.S. equities, AI and other areas. He said that outflow looked cyclical to him and may now be reversing. If money starts rotating back into crypto as an asset class, those tokens should benefit.
For Pump and Hyperliquid, he said that coupling runs deeper. Pump is tightly linked to onchain activity and meme coin trading, and its 90-day average revenue has already grown 80%, with the potential to double or triple again. Hyperliquid's RWA market, covering commodities, stocks and indexes, has meaningful volume but not much revenue yet. Its cash cow remains the crypto token side of the business. If capital returns to crypto, both projects stand to benefit most where take rates are highest.
He used Hyperliquid's fee generation to illustrate the speed of that move. About a week earlier, the platform was generating close to $1 million in daily fees. A few days before the podcast, that figure had reached $5 million in a single day. Venice, by contrast, sits inside the larger AI adoption wave, which he sees as a broader tailwind than crypto market strength alone.
Pump: low buyback multiple, durable revenue debate
From a financial and valuation perspective, Barack said Pump still looks very cheap. His comparison was that Pump trades at around 5x buybacks, while Hyperliquid and Lighter trade closer to 30x to 40x.
The market's skepticism, he said, centers on revenue durability. Investors remember OpenSea, where revenue surged and then fell 95%. Barack argued that Pump is different because its revenue has held for more than two years and is still growing.
His framing was blunt: individual meme coins may be volatile, but Pump is effectively the casino for all meme coins. In his view, that makes it a durable business. He said there is nothing irrational about user demand here. Casinos, lotteries and short-dated options are all huge industries with negative expected value, but people participate because of variance. In that sense, he said, Pump is crypto's version of DraftKings or Las Vegas Sands.
He did acknowledge ongoing uncertainty around the split between equity and token value. The team has committed 50% of revenue to buybacks for 12 months, but the market does not know whether that policy will be extended. Even so, Barack argued that for a team trying to build a multigenerational company, abandoning the token would not be in its interest. He said a more reasonable buyback multiple would be 10x to 14x, implying upside of roughly 2x from current levels.
Hyperliquid and onchain 24/7 price discovery
Barack described Hyperliquid as one of the clearest examples in crypto of moving the full financial system onchain. Instant settlement, round-the-clock markets and broad asset migration onto blockchain rails are the core pieces of that story, in his view.
He said new use cases are already emerging. SpaceX, Cerebras, Unitree and other IPO companies have appeared on Hyperliquid for price discovery. He suggested that in the future, bankers deciding where to price an IPO may look at Hyperliquid screens to see what the market is willing to pay. In his telling, that could become a new form of price discovery.
EtherFi: no longer just a staking story
Barack also spent time on EtherFi, which he said was the first venture investment his fund ever made. EtherFi, in his account, has evolved from a staking-focused business into one built around yield products, credit cards and a more mature brokerage-style offering. He praised the team's execution and said users can trade any onchain asset and borrow against positions through the platform.
He argued that the market still anchors too heavily to EtherFi's old identity as a liquid staking or restaking play, similar to how it once valued Lido. EtherFi reached an $8 billion FDV at the height of restaking enthusiasm and then fell as that theme cooled. But the business today is materially different, he said. More than 65% of revenue now comes from newer banking and brokerage lines, such as credit card fees and lending interest, while only about 35% comes from staking yield. The mix is still shifting as the new brokerage side grows faster.
On valuation, he said EtherFi currently trades at around 10x to 15x earnings. For a business where daily credit card volume has risen from $300,000 to $3 million to $4 million, where programmatic buybacks have just started, and where the token is almost fully circulating with no emissions pressure, he sees that valuation as low.
He cited a Blockworks model that assumes growth is cut in half and still arrives at $21 million of buybacks over the next 12 months. His own assumption is $30 million. Applying a 30x multiple would imply a token price above $1, more than double from current levels, and that does not include any rerating for category leadership.
David added that EtherFi resembles the modern startup model: a small team using large technical systems to move quickly. Because tokenized assets already exist, the company can evolve from a neo-bank to a neo-brokerage with little incremental cost. Ethereum, in effect, does a large part of the infrastructure work for them.
Barack agreed. He noted that EtherFi's lending book is still only $20 million and interest income contributes just 4% of revenue. The business is using existing DeFi infrastructure to scale, including a partnership with Aave to run its own Aave v4 instance under an 80/20 revenue split, with 80% going to EtherFi. Compared with traditional neo-banks such as NuBank, where interest income often makes up 60% to 70% of revenue, he said the runway is still large.
The next cycle belongs to applications
Looking ahead to 2026 and 2027, Barack said crypto has already moved beyond what he called the era of "super infrastructure." Enthusiasm around new base chains is no longer the defining force it once was.
He pointed to a shift in revenue mix. Execution-layer infrastructure used to account for more than 95% of crypto revenue, he said. Now applications account for about two-thirds, with execution layers at about one-third. Over time, he expects applications to generate more than 90% of revenue. The most durable tokens, in that framework, will be tied either to applications or to money.
On the money side, he said Bitcoin is not going away. He also mentioned Zcash as an older privacy coin that is attracting long-time Bitcoin holders by reconnecting with crypto's original ideals. He added that Ethereum's potential as money has become more interesting to him again, given what he described as Bitcoin's quantum risk and concentrated ownership risk.
On the application side, he said Solana remains the most active blockchain and helped create the conditions for Pump. Even without the level of MEV economics seen before, he called it one of the strongest bets on broader crypto adoption.
His final framework was that, in hindsight, the best entry points in 2026 will likely be found in application tokens that crossed from zero to one and in assets that gain durable recognition as money. The names he listed in that category included Venice, Hyperliquid, Pump, EtherFi, Bitcoin and Zcash.