This article is from:Kyle Samani
Compiled by Odaily Planet Daily (@OdailyChina); Translator: Azuma (@azuma_eth)
Editor’s Note: The man most known for hyping Solana, former co-founder of Multicoin Capital Kyle Samani—who recently made a high-profile exit from the space—is back!
Last night, Kyle Samani posted a long thread on his personal X account. In it, Kyle Samani once again demonstrated his persuasive, non-derogatory “call-to-action” style, using efficiency—a weakness in decentralized narratives—as a starting point to explain how Solana’s current flagship PropAMM will match or even surpass the efficiency of traditional centralized models, arguing that PropAMM is one of the most significant innovations in market microstructure in recent years, if not decades.
- Related articles: 《The Man Who Shouted the Most About SOL Has Left Crypto》; 《Did Kyle Samani’s Exit Have a Hidden Reason?》.
The following is the original text by Kyle Samani, translated by Odaily Planet Daily.
PropAMM is one of the most important innovations in market microstructure in recent years, and possibly one of the most important in decades.
To help everyone understand this conclusion, let’s first look at how market makers (MM) quote prices on traditional centralized exchanges (CEX).
Market makers typically colocate physically with the exchange. Each market maker runs its algorithm on a server connected via network cables of uniform length (e.g., 50 meters) to another server running the exchange’s system.
Market makers and exchanges continuously exchange large streams of data. Every time a market maker sends an order to the exchange—whether a limit order, a cancellation, or a market order—the exchange must broadcast this information to all other market makers; those market makers then re-send their own orders based on the new information, and this cycle repeats indefinitely.
Below is a simple diagram.
Now let’s look at how propAMM works on the Solana mainnet.
The beauty of propAMM on Solana is that the blockchain itself directly "hosts" the market-making algorithm. This means the system no longer needs to send billions of messages back and forth between market makers and the exchange—the market-making algorithm runs directly on the same physical machines as the exchange.
The new diagram is shown below. (Yes, only the Solana blockchain is needed!)
It's a common saying in the cryptocurrency industry that decentralized systems, which require communication across global nodes, are inevitably slower (with higher latency) than centralized systems.
But if you look at this question differently, on-chain custodial algorithms may actually have lower latency than centralized exchanges in traditional finance.
Why is this the case? Because the delay required for propAMM to update prices involves only electrons moving within the same physical silicon chip. For example, if the last market order causes a price change in SOL-USD, this information becomes immediately visible to all propAMM and is used to price the next market order. Everything occurs within the same silicon chip, eliminating the need for bidirectional communication between servers.
It should be noted that propAMM does require frequent oracle updates, but this is not an issue and does not change the overall facts I described above.
The most critical point remains that when an exchange — in the above case, the Solana blockchain — directly hosts the propAMM algorithm, market makers' pricing changes in real time within the same physical silicon.
propAMM has become the dominant mechanism for SOL-USDC spot quotes on Solana, with tighter spreads than all major CEXs. I expect this market structure to become the dominant model for on-chain trading this year, including spot, perps, and even prediction markets.
The biggest challenge with propAMM is that there is currently no way to guarantee that takers will always receive best execution, because:
- None of the propAMM algorithms are public (which is actually reasonable, as traditional market-making algorithms are also proprietary);
- Routing trades across multiple propAMMs results in non-deterministic outcomes.
However, this issue is solvable. I expect all related aggregator teams to roll out solutions this year, such as Jupiter and dFlow on the spot side, and Phoenix on the futures side.
The current propAMM has not yet been fully optimized and is still subject to various limitations of the Solana blockchain. This year, Solana will roll out a series of major upgrades that will significantly enhance propAMM's performance, including:
- Higher CU (computation unit) limits per transaction, along with larger trade sizes;
- Higher CU limit per block;
- Alpenglow: Reduced slot time from 400ms to 100–150ms;
- DoubleZero: Reduce global network latency;
- Application-controlled execution;
- Multiple concurrent leaders.
If propAMM on Solana's mainnet was already offering tighter spreads than all CEXs without these upgrades, it’s easy to imagine how powerful its performance will become as these upgrades roll out.