Wyborski opened with an animation of white squares, each one a block holding up to 300 transactions. It was not a simulation and not sped up, he said, but footage captured from Kaspa's test network, which produced ten blocks per second. He gave the talk at Mining Disrupt Miami 2024, held June 24–26, 2024.
Shai Wyborski, one of the original contributors to Kaspa, said early in his remarks that he was not speaking for any foundation or mining company. His target was the conventional wisdom that ASICs make proof-of-work mining centralized. At the time, Bitcoin was trading at $61,272. Today it trades at $82,660.
Why 864,000 Blocks a Day Changes the Math
Wyborski’s central thesis rests on a simple arithmetic comparison. In Bitcoin, the block time is approximately ten minutes, resulting in 144 blocks per day. That puts a hard ceiling on the number of entities that can earn daily revenue from mining: if a network only produces 144 blocks, only 144 solo miners or pools can see that day’s income. To get a steady stream of revenue, he argued, miners have to combine their hash power in pools.
In Bitcoin you have 144 blocks a day, which means you can have at most 144 solo miners or pools that see daily income.
Shai Wyborski · 9:08
Kaspa implements GHOSTDAG, a protocol invented by Yonatan Sompolinsky, who also invented the GHOST protocol; Wyborski stressed that the two are completely different. On the test network it runs at ten blocks per second, or 864,000 blocks a day. In his arithmetic, that means up to 864,000 solo miners or pools can see daily revenue, so small operators can take part without joining a pool. In his framing, how decentralized mining is depends on how often blocks are produced, not on the type of hardware.
The $600 Entry Point for Solo Mining
To illustrate the point, Wyborski turned to the hardware market. He noted that the smallest ASIC available for Kaspa, the KS0, was still a best seller even after larger, more efficient models became available. In most cases, he said, a newer ASIC with a better terahash-per-dollar or terahash-per-watt rate leaves the smaller machines without buyers. On Kaspa, the volume of blocks means even a small device can earn daily.
I think right now you need an investment of about $600 or $700 to get one block a day by solo mining at home.
Shai Wyborski · 10:22
He framed this as a return to the early days of Bitcoin, when anyone could participate. Kaspa, he said, is the second-largest proof-of-work network in the world by capital expenditure, with about half a billion dollars in mining hardware, and still a couple of hundred dollars is enough to be part of it. Even small mining operations, he argued, can meaningfully participate and see profit.
Dismantling the ASIC Centralization Myth
A common critique of proof-of-work is that ASIC mining centralizes the network: ASICs are expensive and specialized, so only large entities can afford them, and hash power concentrates. Wyborski called this conventional wisdom wrong. In his account, the centralization comes from the low block rate of traditional proof-of-work chains. With few blocks, a miner needs a large amount of hash power to see revenue; with many blocks, a small amount is enough.
Now people say that ASIC mining is centralized because you need a lot of hardware and you need to sink in a lot of money.
Shai Wyborski · 8:35
Wyborski further addressed the concern that ASIC manufacturers might hold a disproportionate amount of the coin. He countered that Kaspa’s emission schedule was designed to emulate Bitcoin’s, ensuring that by the time ASICs became dominant, a significant majority of the supply was already in circulation. He noted that in other projects, ASICs often emerged when only 5% or 10% of the supply was circulating, allowing manufacturers to accumulate large holdings. In Kaspa, ASICs emerged when roughly 65% to 70% of the supply was circulating, mitigating this risk.
A Full Node on About $100 of Hardware
Beyond mining hardware, Wyborski addressed the cost of running a full node. High-throughput networks are often criticized for requiring expensive hardware and large amounts of storage, which limits the number of independent validators. Kaspa keeps these requirements low with pruning and a separate mechanism that regulates the growth of the state.
The answer is about 100 bucks.
Shai Wyborski · 17:17
He said people have run full Kaspa nodes on Raspberry Pis at the full throughput of 3,000 transactions per second and ten blocks per second. A validator does not need a $15,000 machine, he said; an unused old PC is enough. By his figures, the network can run at that throughput for more than 1,000 years with less than 100 gigabytes of storage.
The Case for Self-Scaling Proof-of-Work
The final part of the talk concerned confirmation speed. Wyborski separated two properties. The first is reactiveness: confirming a transaction as fast as the network itself allows. He said GHOSTDAG provides this, proven mathematically and seen after years of the network running.
[Kaspa], despite being proof of work, is as reactive as proof of stake systems.
Shai Wyborski · 20:31
The second is adaptiveness. Today's protocols have to assume a bound on network latency with a large margin of error: if a network performs at a latency of two seconds, he said, it has to assume about ten. A protocol called DAGKNIGHT, which at the time existed only in theory, removes that bound. The network would speed up as network conditions improve, and slow down rather than give up security when they get worse. Wyborski suggested that this self-scaling is possible only for proof-of-work, which would let it confirm transactions faster than proof-of-stake.
We can't prove it yet, but we have a lot of reasons to believe that self-scaling is only possible for proof-of-work.
Shai Wyborski · 25:24
Wyborski's case rests on block frequency rather than hardware: the more blocks a network produces each day, the smaller the miner that can earn from it. Mining hardware, pools and decentralization return to the program at Mining Disrupt 2027.