Why Crypto Buybacks Are 'the Wrong Way to Allocate Capital'

The central argument is that early crypto startups often allocate capital poorly when they spend revenue buying back and burning tokens instead of investing in growth. Buybacks may signal that a token matters, but they can also suggest the team sees few ways to expand the business. Founders should explain how investment could grow a market—for example, increasing a stablecoin’s supply from $5 billion to $50 billion—rather than treating token-holder appeasement as the goal. Laura characterized the Pump ICO dynamic as people funding a project to buy its tokens, after which the project uses that money to buy tokens back from them.

Buybacks are not necessarily useful just because they are automatic or announced in advance. The speaker said projects should be able to adjust allocations as market conditions and token value change; a programmatic plan can still be stopped, so investors ultimately have to trust the team. An early commitment to encode buybacks was called a red flag. Projects with substantial funding, especially layer ones, have alternatives such as acquiring companies, technology, or talent.

Hyperliquid was offered as a case where buybacks may not be the best use of capital, despite the project’s current dominant share of the perpetuals market. The speaker floated a possible acquisition of Gemini in the US, partly jokingly and partly seriously, as an example of a different use for capital. Perpetuals are expected to become more competitive and regulated, with Coinbase, Robinhood, and Backpack among possible entrants; stock perpetuals may also be approved.

Lorenzo Valente sees crypto shifting from inventing protocols toward scaling proven products, with finance its clearest product-market fit. That makes value accrual increasingly important, particularly for Ethereum. One speaker estimated that five to seven layer twos account for 99% of Ethereum-related activity, mainly Base and Robinhood, while the base layer captures relatively little activity and few fees. The speaker contrasted Robinhood making several million dollars a day with Ethereum’s base layer accruing roughly $100–$200, though the comparison’s period was unclear. Ethereum, in this view, is caught between scaling layer one to attract applications and building a stronger layer-two franchise with better support and value accrual.

Tokenized assets could make that tension more consequential. Valente expects real-world assets to be a major growth area over the next two to three years, while the speaker said tokenized-equity activity is concentrated on Solana, Base, and Robinhood rather than Ethereum layer one. The guest is bullish that tokenized equities could exceed stablecoins in on-chain supply within three years: stablecoins total roughly $300 billion, while tokenized equities have recently passed 1% of that amount. Early growth may center on a small group of stocks distributed to users in emerging markets and offshore. Perpetuals currently dominate tokenized-equity trading; spot trading will take longer because it requires links to equity infrastructure, custodians, and clearing houses.

Stablecoins have a more established use case: exporting the dollar to emerging markets, and the guest expects them to grow at more than 40% annually over the next few years. Lorenzo said the market is less fragmented than it appears because Tether and Circle benefit from network effects, liquidity, regulation, and high institutional switching costs. He expects Circle to become an broader “internet money” company, but said building a neutral stablecoin network capable of reaching hundreds of billions is difficult. ARK’s $1.3 billion venture fund, ARK VX, has been tokenized through Securitize and can be subscribed to and redeemed on-chain, though it has no on-chain secondary market yet. The effort could improve access across geographies, but remains early; the guest also argued that equities may bring less-correlated yield than crypto strategies based on leverage or looping.

AI is lowering the cost of building and integrating technology, while potential agent payments remain nascent. The discussion cited Cloudflare’s report that more than 50% of internet traffic was bots and its consideration of blockchain nanotransactions, but current activity on MPP and X402, as well as on Solana, Sui, and Aptos, was described as small. Lorenzo estimated layer ones may be two or three orders of magnitude short of the throughput needed for future uses, reinforcing the need for continued scaling.