Blockchain Capital general partners Aleks Larsen and Spencer Bogart said on a recent episode of the Bankless podcast that crypto is moving from an infrastructure-led phase toward an application-led one, with stablecoins, prediction markets and tokenized financial assets at the center of that shift.
The discussion, later compiled by PANews, focused on how abundant low-cost block space is changing where value accrues across crypto. Their argument was straightforward: the base layer buildout has gone far enough that applications can now capture a larger share of revenue and user attention.
Buyback-and-burn remains a live model
The host opened by recalling earlier conversations with Bogart in 2018 or 2019 about Maker’s MKR and its value-capture design. Bogart said they had even considered buying MKR at the time, and added that by 2026 newer projects such as Hyperliquid, Lighter and Venice were still using the buyback-and-burn model first popularized by MKR.
He said critics had spent years debating whether the structure was capital-efficient, but in practice it had been hard to dismiss. Larsen said he had once been much more skeptical, especially from an end-state valuation perspective, because he believed tokenholders would ultimately need direct claim to meaningful cash flow. His view has softened. As he put it, buyback-and-burn works well in the current market.
Bogart tied that durability to the still-unclear legal standing of tokenholders. Unless the Clarity Act passes, he said, tokenholder rights remain murky. In theory, investors may prefer a startup to reinvest cash flow into new growth. In crypto, though, most protocols have not shown that they can expand successfully into entirely new lines of business. For that reason, many tokenholders would rather see a team make a clear commitment to ongoing buybacks and burns than leave capital allocation open-ended.
He also argued that early crypto markets were filled with inconsistent token quality. Serious projects, in his view, often need to prove that seriousness from day one by spending actual money on buybacks and burns. He said the model may not remain dominant five years from now, but for now it is one of the most credible ways to align a team with tokenholders.
A token bear market with unusually positive catalysts
The host raised a split that has become more visible across the industry: some investors say crypto venture is dead, major funds have widened their focus to AI and robotics, traditional financial institutions are showing more interest in blockchain, and many crypto OGs still sound deeply pessimistic.
Larsen said Blockchain Capital tries to zoom out rather than overreact to price swings inside a bull or bear cycle. What makes the current token downturn unusual, he said, is that it is happening alongside the strongest set of positive catalysts the industry has seen in some time. He specifically cited the GENIUS Act, a gradually clearer path for the Clarity Act, and large-scale participation from traditional institutions.
He also pointed to applications that have already escaped crypto’s own information bubble. Prediction markets, including Polymarket, have attracted users who may not care whether the underlying technology is crypto at all. Stablecoins, he said, offer extremely cheap dollar-based cross-border payments and remittances. Both areas have continued to post strong one-way growth even during a bear market. Larsen added that AI absorbed much of the market’s attention over the past year and more, especially after the surge of coding agents and open-source Claude around seven or eight months ago, which pulled focus away from weak token prices.
Crypto is still in the flat part of the S-curve
Asked where crypto sits on an S-curve of adoption, Larsen compared the sector’s trajectory to the commercial internet. The internet began commercializing in 1989 and spent roughly its first decade in experimentation. By 2000 it had hundreds of millions of users, but it was still difficult to use and constrained by bandwidth. The next phase, from 2000 to 2005, was the broadband transition.
He said crypto has just gone through its own version of that transition. Block space has become cheap and plentiful. In his framing, Solana in 2020 was the first chain to show a high-performance scaling path in a monolithic design, while by 2024 Layer 2 networks had become genuinely widespread and Ethereum was also making scaling progress. That, he said, is now the new normal.
But the internet did not immediately break higher after broadband arrived. Larsen said the real upward bend in the S-curve came later, when mobile computing exploded between 2006 and 2010. If Ethereum’s launch in 2015 marks the start of crypto’s clock, the sector is only 10 to 11 years into its development.
He estimated that out of roughly 700 million crypto holders, maybe only 10% are active onchain users. The reason, he said, is that truly consumer-ready tools only became usable in the last two or three years. He listed embedded wallets, social recovery, spending limits and passwordless login as examples of the stack that had to mature before crypto could work for users who do not want to become cryptographers.
That is why he places crypto closer to the internet in 2003 or 2004: after broadband adoption, before the mobile breakout, and still on the flat bottom of the S-curve. In his view, the curve turns upward once fringe applications such as stablecoins and prediction markets move decisively into the center.
Why OGs feel worse even as institutions move in
The host suggested that people who came up in crypto may have been too young and too impatient in 2021, expecting the world to change overnight. That still did not fully explain why long-time participants remain so downbeat.
Bogart described it as a kind of growing pain. When a startup approaches an IPO, he said, early employees often miss the phase when the company felt like a band of rebels. They can struggle to accept what success looks like once compliance, process and scale take over. He compared it to someone who loves a niche band’s early albums, then loses interest once the band becomes mainstream.
Larsen said the same tension is visible at industry conferences, where people in suits are now talking about licensed channels, compliance and access rather than cypherpunk ideals. Still, he said, finance is a deeply regulated sector and it cannot scale without fitting inside rules.
At the same time, he argued that Ethereum and Bitcoin still hold powerful appeal for institutions because decentralization and neutrality offer stronger trust assumptions. The cypherpunk vision has not disappeared, he said. It is simply operating in a quieter and more scalable form as the base network for finance. In his framing, crypto is now upgrading the plumbing of the global financial system, even if that sounds less romantic than earlier narratives.
Application fees overtook infrastructure fees in 2025
The host also noted what he described as a historical first: traditional institutions are leaning into crypto while prices are down and without a manic market narrative to pull them in. At the same time, 2025 and 2026 appear to mark the end of a cycle where infrastructure attracted capital mainly because infrastructure was where the money was going.
Bogart said that in 2019, using Uniswap could cost users a few dollars or even more than $10 in friction. Scarce block space was the industry’s main bottleneck. That pushed too much capital into infrastructure during the hype cycle and helped create today’s environment, where block space is often oversupplied and many blocks go partially empty.
Even so, he said, plentiful low-cost block space remains a basic requirement for developers building applications. The more important shift is in how fees are distributed. In 2021, more than 70% of user-paid fees flowed to the infrastructure layer. By 2025, total application-layer fees surpassed infrastructure-layer fees for the first time.
To Bogart, that marks a real handoff. As transaction costs collapse, value is starting to move up the stack. A healthy ecosystem, he said, should not let the lowest communication layer extract most of the monopoly rent. Breaking that pattern is one of crypto’s core economic promises.
From fat protocols to fat applications
When the host asked whether that meant the old fat protocol thesis was giving way to a fat application thesis, Larsen said yes.
His reasoning was that blockchain networks are supposed to lower intermediation costs and improve efficiency, so the base protocol should not be the point where enormous profits are trapped. The deeper logic, he said, is “thin protocols, large markets.” Even if the take rate at the protocol layer is low, the absolute value captured can still become very large if the network expands the underlying financial market by an order of magnitude.
AI shows familiar patterns, but the pressure is hidden
Larsen said the AI investment cycle shares obvious similarities with crypto’s earlier phases. In crypto, teams could at one point raise multibillion-dollar valuations with little more than a white paper. He sees a parallel in AI labs that command high valuations on the strength of research ambition and elite teams.
The metrics also rhyme. Crypto investors watched testnet TPS and benchmarks; AI investors watch model benchmarks. Token listings on exchanges supplied liquidity and distribution in crypto; hyperscale cloud providers serve as distribution channels in AI.
But one difference stands out, he said. Token prices in crypto are public, liquid and brutally transparent as a sentiment gauge. When a narrative breaks, a token can fall 90% in a month. In AI, the bubble and the downside pressure are still largely hidden in private markets. The correction may show up through down rounds or talent losses rather than the kind of immediate collapse crypto markets display.
On whether the application layer in AI will break out the way it is now doing in crypto, Bogart said absolutely. He pointed to comments from Palantir Technologies CEO Alexander that raw model intelligence alone does not produce the outcomes enterprises want. Someone still has to translate that intelligence into concrete workflows and outputs.
He added that some AI venture investors have recently become anxious about the idea that software has no moat. Crypto investors, he said, find that almost amusing because they have spent the last decade in a fully open-source environment where anyone can fork code at any time and software moats are structurally weak.
Larsen said general model weights are likely to become commoditized over time, but the harness built around them will not. In demanding areas where mistakes are unacceptable, such as semiconductor manufacturing or complex tax audits, applications that combine frontier models, fine-tuning, proprietary enterprise datasets and closed-loop feedback can still build very deep defenses.
Stablecoins as the first successful wave of RWA tokenization
The conversation then turned to tokenized real-world assets, with the host describing stablecoins as the first and most successful generation of RWA.
Bogart said few people realize Blockchain Capital was the only venture firm that invested in all three major stablecoin issuers—Tether, Circle and Paxos—a decade ago. He said the stablecoin market cap is now roughly $300 billion, and that he has better than 90% confidence it will rise to multiple trillions by 2030, “even $2 trillion.”
He said stablecoins were once driven mainly by retail users, but each new growth flywheel now includes institutional participation. Traditional equities, money market funds and U.S. Treasuries can all be pulled onchain because a global, 24/7, programmable settlement layer is simply more capital-efficient.
Bogart stressed that stablecoins are not just a payment product. Their real stickiness comes from what happens after dollars arrive onchain. Most of that money stays, he said, and becomes working capital inside crypto—flowing into lending venues, exchanges and other protocols, where it supports much larger volumes of activity.
He said Blockchain Capital had quantified that effect. Every net $1 billion of new stablecoin issuance creates about $122 billion in onchain economic activity over a year, according to his figures. The same $1 billion also feeds roughly $19 million in recurring annual protocol revenue to downstream onchain protocols.
Tokenized equities could develop in two waves
On tokenized stocks, Bogart said the market could unfold in two waves. The first is access. Global investors, especially those outside the United States, have strong demand for a low-friction, one-click way to trade U.S. equities. The second is composability.
Once a tokenized Apple position exists onchain, he said, multiple lenders and securities-lending protocols can compete in open markets to offer the best collateral terms and yields. That is where capital efficiency becomes much more powerful.
He outlined two competing structures now in the market. One is the X-Stocks model represented by Backed, which has been acquired by Kraken. That structure uses a Cayman SPV to issue debt instruments linked to stocks. Its appeal is that it can move without permissions, without KYC, and circulate freely in DeFi.
Its central weakness is legal and credit exposure. In that setup, the holder owns a claim against the SPV, not an actual share of Apple stock. For large institutions managing tens of billions of dollars, Bogart said, that risk is unacceptable.
The other route is a compliant channel that gives the investor direct ownership of the stock itself. That path, however, requires giving up some of the permissionless qualities that crypto users tend to value.
Traditional finance and DeFi do not need to fully merge
The host asked whether that leaves an unavoidable compromise between traditional finance “suits” and crypto “pirates.” Bogart said no.
He said there is no need to force a full merger. Traditional equities worth tens of trillions of dollars can operate on public mainnet chains in what he called a sidecar model. Those assets may live inside regulatory guardrails, but they can still sit alongside pure permissionless DeFi pools.
In his view, that structure could actually accelerate liquidity for fully cypherpunk systems. Large pools of capital parked in tokenized equities can always be converted into ETH with one click and then deployed in decentralized, permissionless environments.