Hook
In early 2026, as Bitcoin’s fourth halving approaches, Jihan Wu—returned as Bitmain’s chairman—published a memo to the mining community. It warned that demand for next-generation ASIC miners over the next 18 months could exceed current supply by 40%, based on internal capacity models. The statement sent shockwaves through the industry: a call not to rein in production, but to double it. To understand why the world’s largest mining hardware maker would advocate for aggressive expansion just as block rewards halve, we must look beyond simple hash price forecasts. The memo’s hidden premise is that the chip supply chain, not the Bitcoin protocol, is the true bottleneck for network security post-halving.
Context
Bitmain dominates the SHA-256 ASIC market with an estimated 65% share, supplying miners from its Antminer S21 series to the latest S22 Hydro. Competitors MicroBT, Canaan, and BitFuFu trail with smaller footprints. The mining sector is currently in a transition from 7nm to 5nm process nodes, with both efficiency and absolute hash rate at stake. Jihan’s memo implicitly targets two tensions: first, the industry’s cyclical overcapacity panic (which historically cooled ASIC orders during bear markets), and second, the geopolitical anxiety around TSMC’s limited CoWoS packaging lines, which are also required for Nvidia’s HBM-enabled chips. Bitmain’s core thesis is that post-halving, the “difficulty death spiral” is a myth—that the price of Bitcoin will rise enough to absorb the new supply of miners, and that any player who holds back chips will lose market share permanently.
Core Analysis
From the chaos of 2017, we forged a compass. That compass now points to a single truth: capacity, not margins, is the battleground. Let us examine the seven dimensions of this crisis.
1. Technical Process
The S22 series uses TSMC’s N5 (5nm) process, while the competing MicroBT M60 uses Samsung’s 5nm. The key metric is Joules per Terahash (J/TH): the S22 achieves 19 J/TH, down from 30 in the S19. This leap mirrors the DRAM industry’s shift to 1bnm. However, the bottleneck is not logic density but packaging: Bitmain relies on TSMC’s InFO (Integrated Fan-Out) for the ASIC substrate, which shares capacity with AI GPUs. Jihan’s concern about “equipment and timeline” echoes Chey’s earlier lament: the real limit is not the chip design but the number of TSMC CoWoS reactors. The implied message: Bitmain is less worried about 3nm readiness than about securing enough N5 capacity through 2027.
2. Industry Chain
Bitmain’s position in the value chain is that of a fabless designer with massive captive demand from its own mining pools (Antpool, ViaBTC). This vertical integration gives it pricing power upstream (against TSMC) but makes it vulnerable to supply shocks from downstream miners. In the 2022 bear market, Bitmain refused to cancel orders, taking a $500 million write-down. Now, Jihan advocates for the opposite: expand capacity even if short-term demand softens, because the long-term need for hashrate will outpace any interim dip. This is a bet on Bitcoin’s $150K price by 2028. Interestingly, his memo argues that miners should not fret over hash price but should instead lock in ASIC contracts today, treating them as strategic reserves.
3. Capacity and Capex
Bitmain’s latest expansion plan includes a new “super fab” partnership with a major packaging house in Malaysia, aiming to double its annual ASIC output from 1.5 million units to 3 million by 2027. Jihan asserts that “people, equipment, and construction time” will limit actual delivery. The implication: even if all current plans are executed, the network hash rate growth of 30% year-over-year will stall. This is a hidden signal that Bitmain is banking on a supply deficit, which would keep ASIC prices firm. The capital expenditure intensity is extreme: industry analysts estimate Bitmain has earmarked $4 billion for this expansion, funded by its 2024–2025 mining profits. The risk of depreciation is real—if halving pushes undercapitalized miners offline, those ASICs will sit idle. But Jihan’s bet is that institutional miners with low electricity costs will absorb the extra chips.
4. Demand Forecast
Jihan projects that Bitcoin mining demand for new ASICs will grow 50-60% in absolute hashrate terms, driven by price appreciation and the need to replace older S19-class miners that become uneconomical after the halving. This is consistent with historical patterns: each halving cycle sees a surge in efficiency upgrades. He also predicts a 60-100% increase in demand for “premium efficiency” (sub-20 J/TH) miners. The catch: if Bitcoin’s price corrects to $60K, many miners would defer purchases, creating a glut. Yet Jihan argues that the impending supply gap (due to limited chip packaging) will keep ASIC prices high regardless of short-term BTC volatility. He frames the halving not as a threat but as a catalyst for network validity. Trust is not a metric; it is a memory we share—the memory of 2020, when post-halving hashrate grew 80% within 18 months.
5. Geopolitical and Export Controls
Bitmain’s reliance on TSMC (Taiwan) exposes it to US-China tensions. In 2023, the US restricted the sale of advanced ASICs to China-based miners, hindering Bitmain’s direct shipments. To circumvent this, Bitmain now routes through Malaysia and uses SMIC’s 28nm legacy nodes for lower-end miners. Jihan’s expansion in Malaysia is partly a “de-risking” strategy, mirroring SK Hynix’s US factory. However, the US may extend restrictions to include any chip with hash rate over 100 TH/s, which would cripple Bitmain’s flagship S22. His call for “capacity expansion” is also a political move: by making ASICs abundant and affordable, he hopes to reduce the incentive for countries to impose export controls. It’s a gamble that supply-side abundance can counteract state intervention.
6. Competitive Landscape
| Company | Market Share H1 2026 | Key Strength | |---------|----------------------|--------------| | Bitmain | 56% | Vertically integrated, owns Antpool | | MicroBT | 24% | Superior power efficiency in M60 series | | Canaan | 12% | Low-cost manufacturing, small backlog | | Others | 8% | Niche immersion-ready rigs |
The battle is for TSMC’s N5 allocation. MicroBT also uses TSMC but with a smaller budget. Canaan uses Samsung, which has slower turnaround. Jihan’s memo directly targets MicroBT: “To those who prefer efficiency over volume, volume wins.” He is signaling that Bitmain will outbuy its competitors in wafer starts. The risk is overcapacity: if demand softens, Bitmain will be stuck with billions in inventory. But he bets that the network’s need for resilience will create a stable floor.
Contrarian
The contrarian view, and one I hold from auditing 15 ICO whitepapers that overpromised capacity, is that Jihan’s prediction rests on a single point of failure: Bitcoin price must stay above $90K for 36 months. If the post-halving correction is sharper than expected, the so-called “supply gap” will turn into a glut. Moreover, his assumption that difficulty will rise linearly ignores the possibility of a mining capitulation that resets the network. His memo downplays the risk that the halving could actually reduce total hashrate if inefficient miners fail in waves. This is a blind spot: he sees only the demand side, not the self-reinforcing cycle of miner closures. I have seen this hubris before—in 2017, when ICOs promised trustless scaling but neglected the human cost of congestion. Trust is not a metric; it is a memory we share. The memory of 2022, when overleveraged miners sold their rigs at 30 cents on the dollar, should caution against blind expansion.
Takeaway
The industry’s next five years hinge on whether Jihan’s supply-side evangelism is prophetic or desperate. If he is right, Bitmain will have secured a monopoly on the post-halving hash rate wave. If wrong, its balance sheet will be scarred by ghost factories. The deeper lesson is that the Bitcoin network’s security is increasingly a function of semiconductor physics, not just economic incentives. As we approach 2027, the question is not whether we can have enough chips, but whether we can remember why we need them in the first place—to maintain a ledger that is truly immutable, not merely expensive to attack. From the chaos of 2017, we forged a compass. That compass now points to the need for algorithmic soul, where human agency over supply chains matters as much as cryptographic proof.