Update on ASIC Manufacturer AI Oracle Test

QBT has found a workaround to accelerate live testing of its Method C AI Oracle on the ASIC manufacturer's mining rig (the loan was announced 6 March 2026) and now expects to receive earlier preliminary results. This shortens live-testing timelines and represents a modest operational catalyst for the AIM-listed company, although no performance metrics or specific new dates were disclosed.

Analysis

If a pathway emerges to run specialized AI tasks on bitcoin-style ASIC hardware at materially lower incremental energy or capex per inference, the immediate winners are the public ASIC OEMs and the secondary-market brokers that inventory end-of-life rigs. That outcome creates a two-tier compute market: high-margin GPU/cloud for large models and a commoditized ASIC layer for predictable, narrow AI oracles — expect a durable arbitrage opportunity for providers who can certify performance per watt at scale within 3–9 months.

The key risks are binary and scaling-related. A single-rig proof that requires bespoke workarounds is unlikely to change procurement cycles: vendors need consistency across thousands of units and validated software stacks; absent that, commercial demand will lag by 6–18 months. Regulatory and macro crypto cycles are second-order but important — any jurisdictional clampdown on ASIC movement or a crypto price crash could blow out the economics of repurposing and flood the market with low-cost supply, reversing the thesis quickly.

For investors, treat this as an idiosyncratic hardware-arbitrage theme layered on top of the broader AI infrastructure secular. Trade it with concentrated, time-boxed option exposure to public ASIC/crypto hardware names while keeping core long exposure to incumbent GPU leaders that capture most high-end AI dollars. Watch for credible third-party benchmarks (throughput/W/inference and validated software stack) and multi-rig replication as the catalysts that would re-rate OEM specialists within 3–12 months.

Key Decisions for Investors

• Long Canaan Inc (CAN) equity 6–12 months: small tactical position (1–2% NAV). Rationale — highest leverage to any commercial repurposing of ASIC inventory. Target +40–80% if third-party benchmarks appear; stop-loss -30% to limit downside if tests fail to scale.
• Buy NVDA Jan 2028 LEAP calls (or 12–24 month equivalent) as core exposure to AI infra. Rationale — even if ASIC niche wins low-end inference, NVDA retains >80% of high-end training/inference margin. Position size 2–4% NAV; asymmetric payoff of 3:1+ if AI adoption stays on current path.
• Pair trade: long CAN vs short HUT.TO (Hut 8) 6–12 months. Rationale — CAN gains optionality value from ASIC repurposing while miners like HUT are exposed to BTC price and potential secondary-market oversupply. Aim for 30–50% gross spread capture; keep notional balanced and cut if BTC volatility re-prices miners independently.