Quoting because I appreciate Monk raising this and the healthy discussion around it
I agreed retail doesn't care about 50ms vs a couple hundred ms on fills. If we're being honest, they don't really care about execution quality for the most part. Robinhood's wide quotes, Phantom's ~75bps take rate, and Coinbase's ~100bps on retail products are all the proof you need
So the question I'm raising isn't where does retail flow live. It's whether there's a fundamentally better way to do global price discovery for all assets across the entire finance stack
Historically, every major trading venue has operated with first come first serve, which structurally advantages whoever sits closest to the matching engine. As systems became more preformat, exchanges started handing out identical cable lengths inside the datacenter to neutralize latency edges within the building. Once that was solved, the race moved outward capturing market moving information anywhere in the world and racing it back to the one matching engine as fast as possible. Private fiber, microwave towers, custom NICs, custom silicon, the entire HFT industry is a monument to this race and It's been standard practice for decades
My point isn't Solana has retail or Hyperliquid has perps. It's that we may have stumbled onto a genuinely new way to do price discovery, one that finally drags the last corner of the global economy onto the internet, with better execution for the trillions of notional trading volumes across the world
Multiple Concurrent Leaders (or Proposers) captures trade worthy information (alpha) at the edges and collapses the time it takes for that trade to be synthesized. Instead of racing data back to one matching engine in NJ or Tokyo, leaders sit in every major financial hub, ingest local information where it originates, and synchronize globally. The physical distance and time in latency that the trade data needs to travel drops dramatically
It's the same reason Amazon built datacenters around the world, the closer you are to the user, the faster the response. You can't beat physics, so you move the compute to the edge. Markets are no different. The world is big, and the events that move prices are increasingly global. Pinning price discovery to one datacenter in one jurisdiction is an artifact of the preinternet era
My belief is that this is the first-principles correct solution for global price discovery, one that doesn't structurally advantage any single jurisdiction and gives every participant equal access to the state of the world
I'll be the first to admit, if you ask 99 out of 100 market makers whether they want this, they'll say no. It's unproven. It's different. It breaks the mental model they've optimized against for their entire careers
But every major technology shift has a skeuomorphic phase where we just copy the old thing on new rails. I think colocation style onchain exchanges are that phase. The next era is blockchain native designs that actually leverage what the underlying architecture enables pushing finance to move at internet speeds with better price discovery across trillions of volume
I know its strange, unproven, and new, but it's exactly why I believe its a worthy shot on goal. Time will tell if it matters or not.
Some thoughts on Solana v Hyperliquid for perps.
Really loving the recent discussion around Solana perps that
@LoganJastremski and others have been sparking.
Some well-intentioned and genuine pushback.
How do you win over retail flow for this use case?
Ultimately retail flow does not care about philosophical advantages - a truth that worked in your favor when you were competing against Ethereum.
They don’t care about the geographic distribution of your leaders. They, quite frankly, don’t even care about 50ms vs a couple 100ms of latency on their fills. Ask any genuine taker on HL today whether they think the product is inadequate.
A lot of the arguments for why Solana will win perpetuals stems from a perceived distribution advantage, this assumption that Solana has all the retail users already and they just need to offer them perps.
When I first started writing about market structure and engineering challenges on Solana (in the context of enabling perps), I thought that was true.
Back then, all the relevant retail users were on Solana and only the perpetuals power users were really on Hyperliquid.
At this point, that advantage is nonexistent. Hyperliquid houses the strongest and most valuable retail flow in crypto and any onchain user that is even curious about perps has migrated some funds over.
All of the Solana spot trenchers are now Hyperliquid perps traders and the speculative activity that still persists on Solana is no longer as relevant to this equation. The current iteration of the trenches is a shell of its former self.
With MCP it’s looking like some of the market structure issues are going to be addressed. It’s still a rather complex solution but that is obviously what’s required for a more decentralized, general purpose chain.
But if that’s going to get you to (roughly) product parity with Hyperliquid I’m not sure that’s going to be enough, especially considering Hyperliquid itself is a rapidly iterating and improving product.
Just not sure where Solana fits into the perps equation when it’s apparent now that this is an incredibly finicky product to perfect and longer standing, app specific solutions exist.
These third party teams on Solana will have to rebuild retail flow from scratch. I’ve been trying to steel-man the Solana perps thesis but am just failing to see why any perps trader today would choose to go back to Solana for perps when Hyperliquid exists.
Maybe co-location is just not the doomer issue that some people say it is and it’s just about building the best infrastructure for the best product in a world of crypto regulatory capture.
Either way, it’s a worthwhile effort to work towards and would love to see some killer perps teams on Solana. Just putting on my investor hat here.