Solana’s 100M CU Upgrade: Scaling Without Soul?

On a Tuesday morning in late July, I received a Slack message from a young developer I’d mentored during the 2022 bear market. "Michael, they finally did it. 100 million compute units per block. I can build that on-chain order book I dreamed about." His excitement was palpable, even through text. But as I read the Solana Foundation’s announcement—that the mainnet had quietly raised the block compute unit limit from 60 million to 100 million, a 66% capacity increase—my mind drifted not to the technical feat, but to the faces I saw last year at “Rebuild Chicago.” The traders who lost their life savings to sandwich attacks. The DAO members who discovered their “community” votes had been decided by three whales. The visionaries who trusted that scaling would bring fairness, not just speed. Code without compassion is cold. Solana’s architecture is a marvel of engineering—Proof of History combined with a single global state machine that promises 400 millisecond block times and transaction costs under a penny. The compute unit (CU) is its measure of computational work, analogous to Ethereum’s gas. Each block has a maximum CU cap, and until this week, that cap was 60 million. The change, proposed in SIMD-0286, raised the cap to 100 million. In theory, the network can now process 66% more smart contract operations per block. In practice, this is a parameter tweak—not a consensus upgrade, not a sharding solution, not a new virtual machine. It is, in the words of one core contributor, “a dial we can turn.” But dials are never turned in a vacuum. From my experience co-designing governance at UnityDAO in 2020, I learned that every parameter change carries hidden power shifts. Quadratic voting didn’t work because it was mathematically elegant; it worked because we paired it with 42 monthly community calls where people actually discussed proposals. When you increase block capacity, you don’t just create room for legitimate DeFi transactions. You create room for complex MEV strategies—arbitrage, liquidations, and especially sandwich attacks that extract value from everyday users. Solana’s MEV ecosystem, dominated by Jito and a handful of searchers, already accounts for a significant portion of block space. With 66% more room, those searchers can pack more profitable bundles, leaving less room for the simple transfer or swap of a retail user who can’t compete on priority fees. In my days at UnityDAO, we saw a similar phenomenon: when we increased the proposal submission fee to reduce spam, whale proposals surged while small-proposer participation dropped. Parameter changes never affect all actors equally. The technical case for the upgrade is straightforward. Solana’s most demanding protocols—perpetual exchanges like Drift, order books like Serum (now OpenBook), and aggregators like Jupiter—have been pushing against the CU ceiling. Complex transactions that involve multiple swaps, or transfers, or AMM interactions sometimes fail due to hitting the per-block limit. Raising the cap gives them breathing room. And the SIMD process itself is healthy: a formal improvement document, discussion on the forum, validator signaling, and eventual activation. As a DAO Governance Architect, I appreciate that level of process. But I also know that Solana’s validator set, while distributed geographically, remains small in number relative to the network’s economic footprint. Of roughly 1,900 validators, the top 20 control over a third of the stake. Those validators had a direct say in this upgrade—and many of them also run MEV engines. It is rational for them to vote yes: more block space means more fees. But it is also rational to worry about the governance feedback loop where those who benefit from the change are the ones who approve it. Let’s talk about the hidden costs. Every additional compute unit per block increases the block size and the time needed to propagate it through Solana’s Turbine protocol. While the network has successfully handled larger blocks before—witness the NFT mint frenzy of 2022 that nearly choked the chain—this is not a free lunch. Validators on marginal hardware may find themselves unable to keep up, leading to a slow creep toward centralization: only those with top-tier servers and low-latency connections can effectively participate. I recall a conversation during my 2017 Ethical Ledger workshops, where a retail investor asked me, “If the network gets faster, won’t only the rich be able to run nodes?” I didn’t have a good answer then. I still don’t. The upgrade doesn’t change Solana’s hardware requirements explicitly, but it raises the implicit bar for being a competitive validator. The network’s true throughput is bounded not by CU limits but by disk I/O and network speed. Pushing the dial doesn’t change physics; it redistributes the cost of participation. Now the contrarian angle: perhaps this upgrade is evidence not of Solana’s strength, but of its reluctance to confront deeper structural issues. The network has long suffered from bouts of congestion—most famously the failed Candy Machine mints and the 2023 spam attacks. Each time, the response has been to tweak fees or increase limits rather than redesign the fee market or introduce fair ordering. Ethereum, for all its slowness, has EIP-1559 and a robust base fee mechanism that dynamically adjusts block capacity based on demand. Solana’s fee market is a priority fee auction, which—like any auction—tends to favor the wealthy. Raising the CU limit is like adding lanes to a highway during rush hour: it may temporarily relieve congestion, but it encourages more cars (or complex transactions) to pile in, and the traffic jam simply moves to the next bottleneck. I saw this pattern in the DAO world: when we increased the gas limit for voting on UnityDAO, the number of proposals tripled, but the quality plummeted, and the same few voices dominated the discussion. Capacity without structural reform is just more noise. What does this mean for the Solana ecosystem and for the people who depend on it? The immediate benefit is clear: developers building high-throughput applications get a temporary reprieve. Jupiter can execute more complex routes within a single transaction. Marginfi can process liquidations faster. Games can squeeze a few more on-chain actions per turn. But the long-term effect is more ambiguous. If the upgrade attracts more MEV activity, the average user faces higher transaction costs (priority fees) and a greater risk of being front-run. In a sideways market where retail participation is already fragile, that could drive users to alternative chains or L2s that prioritize fairness over raw speed. From my experience leading “Values First” coalition in 2025, I learned that institutional partners care about more than throughput—they care about predictability and integrity. BlackRock’s venture arm didn’t just ask about TPS; they asked about governance transparency and how disputes are resolved. A network that scales by privileging the technically sophisticated will find it harder to build trust with the mainstream. There is also a human narrative dimension. Solana has positioned itself as the “Ethereum killer” on performance, but performance alone does not build a community. The Solana ecosystem is vibrant, no doubt—thousands of developers, billions in TVL, a thriving NFT scene. Yet every time the network solves a scaling problem with a dial-turn, it risks forgetting that real adoption comes from people who feel safe and included. At “Rebuild Chicago,” I met a mother who had invested her savings into a Solana-based lending protocol that got drained in a flash loan attack. She didn’t care about compute units; she cared that the system felt like a casino where the house always wins. Code without compassion is cold. Let me be clear: I am not bearish on Solana. I hold SOL, I build on it, I admire the relentless engineering culture. But as an Evangelist, I believe we must hold the mirror up to our own industry. This upgrade is a technical win, but it is also a governance and ethical stress test. The 66% capacity increase will be absorbed into the network’s bloodstream. If it leads to a healthier, more inclusive ecosystem, I will celebrate. If it merely amplifies the advantages of the already-powerful, I will call it out. My experience at UnityDAO taught me that the best metrics are not “throughput” but “participation rate” and “fairness index.” Let’s watch those numbers in the coming months. For the readers who are waiting for direction: this is not a buy-sell signal. It is a signal of priorities. Solana has chosen to lean into its core strength—raw performance—rather than address the subtler issues of equitable access. That is a defensible strategy, but it comes with trade-offs that the community must acknowledge. If you are building on Solana, ask yourself: Are you adding to the MEV soup, or are you creating protocols that protect the smallest participants? If you are investing, look not just at the TVL charts but at the distribution of fees: Are the gains flowing to the many or the few? The final takeaway is a question: What is the purpose of scaling? Is it to maximize transaction volume, or is it to maximize human flourishing? In my 2020 UnityDAO prototype, we proved that community can thrive even without maximum efficiency—because we built trust. Solana can have both speed and soul, but only if its governance remembers that parameter changes have moral consequences. I’m watching. The network should, too.