The 60% Was Already There: SBI Crypto's Exit Is the Symptom, Not the Signal"

Signal", "article": "The numbers arrived in two columns. On June 30, SBI Crypto Mining Pool reported 16.222 EH/s. By July 31, its telemetry read 0.452 EH/s. A 97% collapse in thirty-one days. The coverage called it a warning about mining pool consolidation.\n\nIt was not.\n\nThe 60% threshold had already been breached before SBI turned off its Stratum servers. On July 20, the top three pools — Foundry, AntPool, F2Pool — accounted for 64.8% of attributed blocks. On July 27, they held 60.8%. The concentration existed as a structural fact before SBI decided to leave. Its exit did not create the danger. It just made the danger legible.\n\nThe code did not change. The difficulty algorithm did not change. The UTXO model did not change. A Japanese financial conglomerate concluded that mining pools no longer justified the electricity bill. That is an economic story, not a protocol story.\n\nWe keep staring at the exit. We should be staring at the distribution.\n\nSBI Crypto is a subsidiary of SBI Holdings, one of Japan's largest financial groups. The pool launched in 2018, selling something rare in the mining landscape: compliance. For miners who wanted distance from Chinese operators and gray-market finance, SBI offered Japanese regulatory hygiene. It never reached the top tier. For most of its operating life, its block share floated in single digits.\n\nCompliance costs money. Japanese corporate governance demands audit trails, legal review, and contingency procedures that a two-person pool in a low-regulation jurisdiction can ignore. SBI carried that weight while competing against operators with a fraction of the overhead.\n\nMining pools are not miners. They are coordinators. Miners connect through the Stratum protocol, submit shares, and the pool assembles candidate block templates. The pool operator decides which transactions enter the template. The pool operator sets payout policy. The pool operator holds the backend keys. Miners supply raw hashpower. This matters because nearly every \"mining pool market share\" statistic is a block attribution statistic — it measures which pool mined a block, not which machine pointed the hardware.\n\nHashrate Index data tracks the collapse. SBI's seven-day average declined from 16.222 EH/s to 5.817 EH/s across July, a 64% monthly decrease. After July 29, SBI stopped producing blocks. Its 24-hour average fell to 0.452 EH/s. Final attributed block share: 0.72%, roughly 6.8 EH/s. The shutdown was orderly, executed in phases.\n\nThe concentration metric climbed in parallel. Foundry led at 26.67% of attributed blocks. AntPool followed at 17.13%. F2Pool at 16.21%. Three pools, roughly 60% of the network's attributed production. Weekly bucket data confirms this predated the shutdown announcement: 64.8% on July 20, 60.8% on July 27.\n\nThe pattern was written into the ledger before anyone published a press release.\n\nNow the dissection. What does the 60% number mean, and what does it not mean?\n\nFirst: block attribution is not hashrate measurement. Hashrate Index counts blocks won. A pool attributed 26.67% of recent blocks probably controls a similar share of global hashpower, but \"probably\" carries real weight. Mining is a Poisson process. Variance exists. One block win shifts a daily reading by a visible fraction. The widely repeated 60.01% figure is a snapshot of a single moment, not a proven state of persistent control. Treating it as a permanent cartel is analytically lazy. I have spent years reading pool charts, and the first rule is distrust the single data point.\n\nSecond: the concentration predates the trigger event. The July 20 reading of 64.8% arrived while SBI still nominally operated. SBI was already a rounding error — 0.72% of attributed blocks. Its exit changed the ranking chart but not the underlying power structure. Analysts who blamed SBI's closure for the 60% number inverted the causality. Consolidation caused SBI's exit, not the reverse. Direction matters because misreading cause and effect produces bad policy responses.\n\nThird: the technical layer under stress is the pool layer, not the consensus layer. Bitcoin's protocol sits untouched. Proof-of-work. Difficulty adjustment. UTXO ledger. All unchanged. The chain kept producing blocks at the expected interval. In the blockchain, truth is coded, not claimed. SBI's Stratum infrastructure shutting down is churn at the service tier, and miners face low switching costs — reconfiguring a miner to a new pool takes seconds. The protocol did not flinch.\n\nThe uncomfortable part is where the pool layer concentrates power. Pool operators select transactions. They determine what enters blocks. They process payouts. They hold administrative control over templates. When three pools control 60% of attributed production, they control 60% of transaction inclusion. Beneath that sits decentralized consensus; above it sits a practical oligopoly. That gap between the ideal architecture and the operating reality is where the risk lives.\n\nThis is not hypothetical. In 2014, Ghash.io crossed 50% of network hashrate and the community recoiled, demanding the pool self-limit. Concentration has been a known vulnerability since Bitcoin's early years. We accepted it then as an anomaly. The SBI numbers suggest we have slowly normalized it.\n\nProtocol analysts distinguish liveness from safety. Liveness — the chain keeps producing blocks — is unaffected by pool concentration. Safety — the chain resists reorganization and censorship — weakens as coordination rises. SBI's exit does not touch liveness. It makes the safety question more urgent, not less.\n\nDuring my DeFi lend-or-die audit work, I learned to distrust elegant interfaces. Compound v1 looked clean. Its interest rate model was mathematically elegant. The edge cases could drain liquidity under specific volatility conditions. SBI's pool interface was similarly polished. It still could not survive the economics. During the Terra-Luna collapse forensics, I traced bridge outflows for six weeks and learned that the consensus layer can hold while the economic layer breaks. Mining pools occupy that same fragile gap: the chain stays honest while the coordination layer concentrates.\n\nLet me run the token-level numbers. SBI's exit removed roughly 0.07% of network hashrate. The network absorbed it with under 2.5% impact — noise in a system that absorbs 30% hashrate swings during migrations. Bitcoin's supply schedule is untouched. Block subsidy: 3.125 BTC post-halving. Fee market: independent of pool operators. The macro token economics do not change.\n\nThe micro economics do. Mining pools charge between 1% and 4% of miner production as fees. Fewer pools means fewer independent fee setters. Foundry, AntPool, and F2Pool accrue pricing power as the field narrows. Miners can migrate by editing one URL, which sounds trivial. It is. But migration only helps when a competitive destination exists. With three dominant price-setters, the choice is often theoretical.\n\nSBI's exit also reflects the post-halving squeeze. The April 2024 halving cut block subsidies from 6.25 BTC to 3.125 BTC. A pool with identical hashrate now earns half the Bitcoin it earned eighteen months ago. Fee income scales down proportionally and the cost side does not. Add Japan's industrial electricity rates — among the highest in the developed world — and the arithmetic turns negative. SBI's decision was not mysterious. It was a business model colliding with halved revenue and elevated input costs.\n\nThere is a regulatory footnote. SBI Holdings operates under Japanese financial supervision, and a mining pool has little strategic value to a conglomerate facing digital-asset headwinds at home. The pool was probably the first asset cut when the budget review arrived. That is how large organizations behave: they prune what does not serve the core business.\n\nThe market structure tells the same story. Foundry's 26.67% leads on institutional credibility. AntPool's 17.13% follows on Asian mining-channel depth. F2Pool's 16.21% holds third as the oldest survivor. The middle tier is churning. Luxor is rising, differentiated by hashrate derivatives and data products. Braiins is falling, respected for open-source software but commercially thin. NeoPool has vanished from recent rankings, which usually means dormant or exiting. This is consolidation in motion: the top firms widen their moats, niche players specialize or die, and the middle gets squeezed.\n\nBehind every rug pull is a pattern of neglect, and behind every pool exit is a pattern of economics. The neglect here was not SBI's operations. It was the industry's acceptance of block attribution as a proxy for truth.\n\nThe hidden variable is migration. SBI's telemetry measured hashrate attributed to SBI's service. As SBI wound down, some miners likely pre-migrated to other pools ahead of the official cutoff. The public charts understate the real redistribution. Block attribution data cannot show where the hashpower went — only that SBI stopped winning. The statistical blind spot is real. Following the hash is not the same as tracking it. Without Stratum-level connection data, the post-SBI map of hashrate remains incomplete. The official charts show where the wood was removed, not where it went.\n\nThere is another underappreciated variable: block template policy. Top pools increasingly differentiate by whether they run Bitcoin Core's default standard template, whether they include Ordinals or BRC-20 transactions, whether they mine empty blocks during fee spikes. These are not technical differences. They are governance preferences expressed through block production. In a consolidated market, the top pools' preferences become the network's de facto transaction policy. Hashrate charts do not capture this power.\n\nMiners are not asking about decentralization metrics. They are asking whether their payout arrives on schedule and whether the pool will still exist next quarter. The SBI closure answers that question for its former clients: migrate fast. The structural lesson is to hold more than one Stratum URL. Loyalty to a pool is a liability. Hashpower is liquid; treat it that way.\n\nNow the contrarian reading.\n\nThe bulls are not entirely wrong. Bitcoin's security model does not collapse because a pool closes. Actual security derives from the difficulty adjustment and the economic cost of attacking the chain. A three-pool world is more fragile in theory. In practice, the three pools have diverging incentives. Foundry wants institutional legitimacy. AntPool wants Asian market share. F2Pool wants longevity. Their interests overlap on survival but split on policy. That friction prevents a clean cartel.\n\nConcentration is also reversible. The 2021 China mining ban reshuffled pool rankings within weeks. Miners are mercenaries. They chase fees, payout frequency, and uptime. If Foundry raises fees, AntPool captures share quickly. Concentration is stickier than it was five years ago, but it is not a locked position.\n\nThere is a structural argument for institutional pools as well. Foundry answers to US regulators. It maintains custody standards. It is accountable to subpoenas. An American operator subject to legal process may be preferable to an anonymous pool in a gray jurisdiction. That conclusion is uncomfortable for Bitcoin idealists. It is nonetheless what the compliance environment rewards.\n\nHere is where the bulls miss. Concentration creates coercion points. A three-pool hashrate layer is a chokepoint that states have noticed. The US government can compel Foundry. It can. That makes Bitcoin's most important coordination function dependent on one jurisdiction's goodwill. Calling this a paper tiger ignores how pools already