High TPS Crypto Presales: What the Throughput Numbers Actually Mean
Every cycle, a fresh wave of high TPS crypto presales lands in your inbox claiming 50,000, 100,000, or “over a million” transactions per second. The pitch is always the same: this chain is faster than Visa, faster than Solana, faster than the laws of physics apparently allow. If you have been around long enough, you already know how this story usually ends — testnet numbers that never reproduce on mainnet, validators that mysteriously all run in the same data center, and a token chart that decouples from the throughput claim about three weeks after launch.
This guide is for retail readers trying to figure out which TPS claims in a presale deck are engineering and which are marketing.
What “TPS” actually measures (and what it hides)
Transactions per second is a deceptively simple number. The honest version of the question is: which transactions, on what hardware, under what network conditions, with what finality?
A token transfer is cheap. A swap on a DEX touching multiple pool contracts is not. A simple ed25519 signature verification is fast; running EVM bytecode against deep state is slow. When a presale deck says “100,000 TPS,” it almost always means the cheapest possible transaction, on a small validator set, in a lab.
For comparison, real public networks under real load look very different. Ethereum L1 processes roughly 15 to 25 transactions per second according to Etherscan’s transaction chart. Solana, despite a 65,000 TPS theoretical ceiling, has historically averaged a few thousand non-vote TPS in normal conditions, with periodic congestion-related slowdowns and outages documented across public dashboards. The gap between the brochure number and the lived number is the entire story.
The scalability trilemma is still real
Vitalik Buterin’s scalability trilemma post is from 2021 but the engineering tradeoff has not been repealed. You can typically pick two of: decentralization, security, scalability. A chain advertising massive TPS is, almost without exception, trading off one of the other two — usually decentralization, by running a small validator set on beefy hardware, or security, by weakening finality assumptions.
When you see a presale deck, the questions to ask are concrete:
- How many validators? What is the hardware requirement?
- Is the consensus deterministic-final, probabilistic, or hybrid?
- What is the time to finality, not just inclusion?
- Has the benchmark been reproduced by an independent third party?
If the answer to the last one is no, the TPS number is a claim, not a measurement.
Common ways presale TPS numbers are inflated
A few patterns repeat across decks. Knowing them is half the defense.
Parallelized empty transactions. The benchmark fires millions of native token transfers between pre-funded accounts with no contract calls. This is the easiest workload to scale. Real economic activity is not this.
Optimistic theoretical ceilings. “Up to” 1,000,000 TPS is meaningless without a sustained measurement. Sui’s documentation, for example, explicitly distinguishes between owned-object and shared-object transaction paths because the throughput differs sharply between them.
Aggregating across shards or rollups. Adding up the TPS of 100 separate execution environments and calling it “the chain’s TPS” is a marketing trick. Cross-shard composability has its own latency cost that never appears in the headline number.
Single-node benchmarks. A single sequencer pumping transactions into a queue is not a decentralized network. It is a database with extra steps.
What to actually check on a presale that pitches throughput
Before you put any money into a high-throughput presale, run through this short list. We expand on the broader process in our upcoming presales guide and in our presale scoring methodology.
- Find the testnet. If there is no public testnet you can hit with your own load, the TPS number is a claim. Period.
- Read the consensus paper, not the pitch deck. If there is no paper, that is the answer.
- Check validator decentralization assumptions. A 21-node permissioned set running on identical AWS instances in one region is not a chain, it is an availability zone.
- Look at finality time, not just throughput. A chain doing 50,000 TPS with 60-second finality is slower for many real applications than a chain doing 2,000 TPS with sub-second finality.
- Cross-reference team and audits. High-performance consensus is hard. If the team has no prior distributed systems track record, scepticism is the appropriate default. See our wallet and infrastructure review approach for how we weight resumes.
- Custody. None of the TPS matters if you lose the tokens before the chain launches. Our self-custody guide covers the basics.
What we could not verify
For most of the high-TPS presales we see pitched in 2026, we cannot independently verify:
- Sustained mainnet throughput under adversarial conditions
- Validator hardware diversity
- Reproduced benchmarks by neutral third parties
- Whether the listed advisors are actively involved or just on a website
If a project objects to that list being made public, that itself is a data point.
Where TPS does and does not matter for token price
This is the part presale marketing never says out loud: throughput and token price are weakly correlated at best. A faster chain does not automatically have a more valuable native token. Demand for blockspace, fee burn mechanics, validator economics, and supply schedules drive price far more than the raw TPS headline. Several chains in the top 50 by market cap have lower throughput than chains ranked 200th. The market is not paying primarily for transactions per second.
If you came here because a presale told you its TPS number justifies the valuation, that is the marketing bridge being sold to you. The valuation needs its own justification.
Honest summary
High TPS crypto presales are mostly selling a benchmark, not a network. Throughput claims in pitch decks are almost always measured under best-case conditions that will not survive contact with real users, real contracts, and real adversaries. Treat any number above the current Solana mainnet average as a marketing figure until proven otherwise on a live, decentralized network with independent reproduction. And remember: even a genuinely fast chain is not, by itself, a reason to buy a token at presale prices.