Moore’s Law

Moore's Law is the observation, first made by Intel co-founder Gordon Moore in 1965, that the number of transistors that can be economically packed onto an integrated circuit doubles at a regular cadence. Moore originally projected a doubling every year, then revised it in 1975 to roughly every two years, the figure most people quote today. It is not a law of physics but an industry-shaping trend — part observation, part self-fulfilling prophecy, since chipmakers planned their roadmaps around hitting it — and it held for more than fifty years, driving the relentless cost and performance gains behind all modern computing, including Bitcoin mining hardware.

What it really predicts

Crucially, Moore's Law is about transistor count and the economics of cramming more devices onto a die, not directly about speed or power. For most of its run, each doubling roughly halved the cost per transistor, which is what made successive generations of mining ASICs cheaper per unit of compute. The speed-and-power side of the story belonged to its companion, Dennard scaling, which said smaller transistors could also run faster at the same power density — and which broke down in the mid-2000s, well before transistor doubling itself slowed. Since then the industry has kept density climbing through structural invention rather than simple shrinking: the planar transistor gave way to the FinFET, and FinFETs are now yielding to gate-all-around designs, each generation buying back electrostatic control that plain scaling could no longer provide.

Why miners feel it slowing

For years, mining efficiency improved partly by riding Moore's Law to denser, cheaper transistors — each ASIC generation delivered dramatic joules-per-terahash improvements largely because the underlying process node did. With the cadence stretching out and leading-edge wafers costing more at every node, gains now lean more on packaging, voltage tuning, and circuit design than on raw transistor doubling. The result is visible across recent hardware generations: efficiency improvements have become incremental rather than generational leaps. Network difficulty still climbs as deployment expands, but the days when a two-year-old machine was hopelessly obsolete are fading.

The repair-economics dividend

This slowdown quietly rewrites mining economics in favor of the small operator. When efficiency doubled every generation, repairing an old machine rarely made sense — the replacement was so much better that broken hardware was scrap. Now, with new machines only modestly more efficient than the fleet they replace, a repaired hashboard keeps earning for years, resale markets stay liquid, and the economically rational lifespan of a miner stretches far beyond its warranty. A machine's useful life is increasingly decided by maintenance and electricity price, not by silicon obsolescence. That is precisely the world where repair skills, spare parts, and efficiency tuning through underclocking compound in value — the end of easy scaling is, from the repair bench's perspective, the beginning of hardware that is finally worth keeping alive.

Moore's Law and decentralization

There is a decentralization argument buried in the economics. When transistor scaling ran hot, the newest node conferred a decisive advantage, concentrating viable mining in whoever could buy the latest silicon first — capital advantage compounded every cycle. As scaling matures, the efficiency gap between the freshest ASIC and a well-maintained older fleet narrows, and competitiveness shifts toward the things small operators can actually win: cheap or stranded power, waste-heat reuse, uptime discipline, and repair skill. A home miner heating a workshop with a previous-generation machine is closer to the efficiency frontier today than at any point in mining's history. The slowing of Moore's Law is often narrated as an ending; for distributed hashrate — and for the repair benches at D-Central that keep older silicon alive — it reads more like a leveling of the field.