Solana co-founder Anatoly Yakovenko has named five major applications as higher priorities for network optimization than validators, developers, market makers, or core engineers, arguing that products holding substantial assets and handling retail trading volume deserve the most attention. Summary Five apps topped Yakovenko’s list of groups that Solana should optimize its network for. He credited the Ethereum Foundation with serving Aave’s need for reliable transaction access. His post responded to criticism of Ethereum designers’ personal experience using the network. The comments contained no specific upgrade, funding allocation, or change to validator participation rules. Anatoly Yakovenko made the comments in an Oct. 23 X post, naming Phantom, Pump.fun, Fomo, Jupiter and Kamino while responding to criticism of how Ethereum’s protocol designers use their own network. Products with a ton of aum and a ton of retail volume are the two most important groups to optimize for. Phantom/pump/fomo/jup/kamino validators or devs or market makers or core engineers. FWIW, eth foundation has done a decent job at optimizing for aave, because the… https://t.co/DCpN0XaM1Z— toly 🇺🇸 (@toly) October 23, 2026 Solana’s largest apps top Yakovenko’s priority list Yakovenko identified assets under management and retail activity as the two measures behind his preference. His post did not provide figures for either measure or rank the five named products against one another. “Products with a ton of aum and a ton of retail volume are the two most important groups to optimize for,” he wrote. The co-founder then placed the five applications ahead of validators, developers, market makers and core engineers. His wording addressed whose needs should guide optimization, without specifying a technical proposal or a timetable for implementing one. Phantom, Pump.fun, Fomo, Jupiter and Kamino were examples in his argument. The post did not announce an agreement with those teams, give them governance authority, or identify a network setting that would change on their behalf. In a follow-up visible beneath the post, he described the priority user in more general terms. “The most important user is the plumber, institutional or retail plumber.” Ethereum criticism prompts praise for Aave’s reliability needs The discussion began with another poster’s argument that many people designing Ethereum’s protocol do not use it themselves. The author argued that designers should depend on their own product and face losses when it fails. Quoting that criticism, Yakovenko gave the Ethereum Foundation qualified praise. He said it had done a decent job optimizing for Aave because an inability to get transactions included during an outage or congestion would break the lending platform. Application fees featured in an earlier Robinhood debate Yakovenko previously argued that a company could build on Solana, subsidize transaction costs and charge customers through its own interface. As crypto.news previously reported on Sep. 6, his position drew a response from Offchain Labs co-founder Steven Goldfeder over Robinhood’s decision to use Arbitrum technology. Goldfeder defended Robinhood’s ability to retain chain revenue, saying it could keep roughly 90% under the Arbitrum Expansion Program. He contrasted that arrangement with an application paying to use Solana. The reported arrangement sends 10% of net protocol revenue to the Arbitrum ecosystem, divided between an 8% allocation to its DAO and 2% for its Developer Guild. The percentage concerns net revenue after relevant network expenses, including Ethereum data-posting costs. The Robinhood exchange concerned a U.S. brokerage’s infrastructure and revenue model. Yakovenko’s Oct. 23 post did not announce any change to product access for U.S. customers. Alpenglow preparations set out separate duties for app teams and validators Earlier Solana upgrade coverage documented different requirements for application developers and the operators supplying their infrastructure. On Sep. 26, Alpenglow reached devnet, following its testnet transition, giving application teams access to the new consensus behavior before a mainnet handoff. The Foundation’s guidance, described in that coverage, said applications that only submit transactions and read balances would not need a migration. Services processing blocks or tracking validator activity faced changes to how they read network data. An Aug. 14 account of the Alpenglow finality target documented an objective of roughly 150 milliseconds, compared with about 12.8 seconds under TowerBFT. The coverage distinguished consensus finality from an initial indication that a transaction landed. Subsequent reporting on the validator fee changes described the Foundation’s plan to replace individual onchain voting fees with a flat 0.8 SOL admission charge per epoch, burned in full. Operators must fund that charge above their vote account’s rent-exempt minimum. The same guidance makes payment one epoch in advance. A validator unable to pay loses permission to vote or produce blocks in the following epoch. The operator documentation described in the October coverage qualifies the finality target for nodes outside the admitted validator set. Those nodes normally learn the result from certificates included in a later block, potentially adding a produced-block interval, with longer waits if blocks are skipped or delayed. For the earlier testing stage, the September coverage distinguished the two public networks. Devnet lets application teams check software using tokens without real value, while testnet lets validators and infrastructure operators put network software through more demanding conditions. Anza publicly announced the successful devnet transition on Sep. 25, one day after testnet completed its switch.
