The key thing about @PhoenixTrade isn't whether fees are cheaper to trade. It's the engineering feat they're pulling off. Phoenix is a fully onchain orderbook DEX running on a general-purpose VM. That design lets it compose atomically with arbitrary smart contracts, which means DeFi can plug into it in a far more seamless way than @HyperliquidX's appchain + general VM sidecar model can. That composability is the whole point. Phoenix isn't just a venue, it's infrastructure the rest of Solana DeFi can build on directly. And that's exactly why it maps so cleanly onto the @solana thesis more broadly: one general-purpose chain where execution and composability live in the same place. On this BTC example, Phoenix already tracks Binance materially tighter than Drift, holding 76% of hourly observations within ±5 bps versus Drift's 57%, while also avoiding sharp dispersions that weighed on Drift's performance. Based on this analysis, Phoenix still carries a small positive basis, but the gap is much narrower and impressively close to app-specific venues like Hyperliquid and @pacifica_fi. That convergence is the signal to watch: if Phoenix's execution continues to tighten, it strengthens the case that Solana-native perps can approach the quality of more centralized or app-specific venues. And note that we are still early. This is before @anza_xyz's Alpenglow and MCP, and before @jito_sol's BAM perp plugins. I've always believed the chain that combines a general-purpose environment with application-specific execution quality will ultimately see runaway network effects. Phoenix is the player to watch on this thesis right now: if Solana can support high-quality native @perps, it becomes much easier to imagine the same architecture extending into other latency-sensitive markets, from prediction markets to more exotic onchain derivatives.