The progression from general-purpose computing to specialised hardware happened quickly and permanently changed who could participate.
The initial period
Mining on ordinary computer processors, accessible to anyone.
Which lasted a short time before more efficient approaches emerged.
Individual participation with consumer equipment was viable during this period only.
Graphics hardware
Substantially more efficient for the parallel computation involved.
Which displaced processor mining rapidly.
Gaming hardware shortages attributable to mining demand occurred repeatedly during later cycles.
Programmable logic
An intermediate step offering better efficiency than graphics hardware.
Which had a brief window before purpose-built chips arrived.
The technical barrier to programming these limited adoption.
Purpose-built chips
Silicon designed for one function, orders of magnitude more efficient.
Which ended participation with general-purpose hardware entirely.
Development requires substantial capital and access to advanced manufacturing.
Manufacturer concentration
A small number of firms design and sell this equipment.
Which produced concerns about manufacturers mining with hardware before selling it.
Allegations and disputes on this point have recurred.
Resistance attempts
Some networks selected functions intended to prevent specialised hardware.
Which succeeded temporarily and generally failed over time.
Memory-hard functions raised the barrier without eliminating it.
The pooling consequence
Individual miners joined pools to smooth income.
Which concentrated block construction among pool operators.
Protocols allowing individual miners to build their own templates address this and have seen gradual adoption.
What it means for the network
Security is provided by an industrial sector with concentrated hardware supply, which is a different structure from the original design's assumption.
Manufacturing constraints
Advanced chip fabrication capacity is limited and allocated among competing demands.
Which affects lead times and cost for specialised mining hardware.
Access to leading-edge processes has been a competitive factor between manufacturers.
Efficiency progression
Energy per unit of computation has fallen by orders of magnitude across hardware generations.
Which is documented in published specifications over time.
The rate of improvement has slowed as the technology matured.
Second-hand markets
Older equipment finds use where power is cheap enough.
Which extends effective life beyond the leading edge.
Machines have operated profitably for years in low-cost locations after being displaced elsewhere.
Consumer hardware effects
Graphics card availability and pricing were affected during periods when that hardware was profitable to mine with.
Which ended for the largest network once specialised chips arrived and for others after consensus changes.
What this means for participation
Individual mining on general hardware is not viable on major networks, and pooled participation with specialised equipment is the realistic route.
Immersion and cooling innovation
Liquid cooling permitting higher density and improved efficiency.
Which requires higher capital expenditure.
It also extends hardware life by reducing thermal stress.
Firmware optimisation
Custom firmware tuning efficiency beyond factory settings.
Which is a meaningful margin improvement at scale.
It carries warranty and reliability trade-offs.
Supply chain concentration
Manufacturing, assembly and distribution are concentrated geographically.
Which creates exposure to trade policy and shipping disruption.
Tariffs and export controls have affected equipment availability in specific markets.
Effects on other networks
Consensus changes on one major network released substantial general-purpose hardware.
Which affected difficulty and profitability on remaining networks using that hardware.
The structural outcome
Network security now depends on an industrial supply chain with a small number of participants at each stage.
What it means for the design assumption
The original conception assumed participation with ordinary hardware.
Within a few years that became impossible, and network security moved to an industrial sector with concentrated manufacturing and geographic clustering.
Whether that matters is genuinely debated; that it happened is not.
Where participation is still possible
Pooled mining with current-generation hardware where power costs permit.
Smaller networks with hardware-resistant functions, at correspondingly smaller scale.
Solo mining on major networks is effectively a lottery ticket rather than a business.
The honest calculation
Power cost, hardware efficiency and expected difficulty growth determine everything, and the arithmetic is straightforward.
The wider industrial picture
A network's security now depends on chip manufacturing, energy markets and shipping logistics.
None of that was in the original design's assumptions, and all of it is now load-bearing.
The relocation episode demonstrated both the fragility and the resilience of that arrangement simultaneously.
Energy sourcing
Operations locate where power is cheapest, which is frequently stranded or curtailed generation.
Which is the basis of the industry's grid-benefit argument.
Whether that argument holds depends on whether operations actually curtail when asked.
Decommissioning
Obsolete hardware becomes electronic waste in substantial quantities.
Which is a documented and growing disposal issue.
Recycling and reuse markets exist and absorb a fraction of the volume.
A closing note
Within roughly five years, mining went from something anyone could do on a laptop to an industrial activity requiring purpose-built silicon, cheap power at scale and access to constrained manufacturing.
That transition was not planned, was not prevented, and permanently changed who secures the network.
Everything downstream — pool concentration, geographic clustering, manufacturer influence — follows from that single transition.