KEEL · Keel Infrastructure Corp.

Current Thesis

KEEL is Bitfarms reborn the old BITF Bitcoin miner that redomiciled Canada→US (Delaware parent), rebranded, and replaced BITF on Nasdaq + TSX on 2026-04-06. The narrative leg is the BTC-miner→AI/HPC-datacenter pivot: turning a 2.2 GW grid-connected power pipeline (Panther Creek PA, Sharon PA, Moses Lake WA) into long-duration AI lease cash flow. The stock ran ~8x off its $0.70 52-wk low to a $6.60 high. That leg is now cracking, not accelerating: on 2026-06-04→05 the company priced an upsized $400M convertible (up from $350M), and the stock fell from $6.10 (06-01 close) to ~$5.13 (06-05). The re-rate still hinges on a single unwon binary the first signed HPC/AI tenant lease, of which there are zero as of 2026-06. The financing confirmed the funding gap the bears flagged while the lease test remains open, so the tape has flipped from breakout to distribution.

Bull Case

• Identity flip is real. Legal Delaware redomicile + rebrand completed 2026-04-06, replacing BITF on Nasdaq + TSX the cleanest pivot execution in the ex-miner cohort.
• Convert structure is shareholder-friendlier than feared. The $400M notes (priced 2026-06-05) carry a 1.250% coupon, mature 2032-01-15, and convert at ~$7.41 a 25% premium to the 2026-06-04 $5.93 close. Cheap, long-dated capital with premium-struck dilution beats an at-market equity print.
• Pipeline scale + real interconnect. 2.2 GW pipeline; 341 MW energized and 430 MW secured as of early June; Panther Creek alone is 350 MW across 336 acres with headroom past 500 MW, backed by a $300M Phase 1 project-finance facility, energization targeted year-end 2026.
• Cohort tailwind is live. Peer Applied Digital signed a $7.5B, 15-yr, 300 MW AI lease (Apr 2026); IREN/WULF/CIFR/CORZ are re-rating from compute-commodity to AI landlord. KEEL is the late mover with the most pipeline-per-market-cap optionality if it lands a tenant.
• Real revenue base. TTM revenue $218.6M (+67.4%); FY2025 $229.3M (+72% YoY). Liquidity ~$533M as of 2026-05-08, which management says funds the three sites through lease execution.

Bear Case

• The lease test is still unwon. The entire re-rate is "lumpy power revenue → utility-like lease stream," and zero HPC/AI leases are signed. The stated goal is three leases by year-end 2026 (one each site); Q2 (ends 2026-06-30) is closing with none announced.
• Financing selloff is the market voting down the funding gap. $400M convert + $58M greenshoe (13-day option from the 06-05 pricing, ~06-18) closing ~2026-06-09 confirms capital need; convert-arb desks short the stock to delta-hedge, which is the mechanical driver of the ~16% drop into the weekend.
• Dilution overhang stacks on a heavy share count. Full conversion is ~54M shares ($400M × 134.9073/$1,000), ~9% on the 603.8M base, more with the greenshoe on top of ~$573M prior LT debt now plus the new notes.
• Cash burn is structural. TTM net loss −$374.3M; FY2025 −$284.5M; latest quarter net loss −$145.4M, EBITDA loss −$96.3M, FCF ≈ −$75M/qtr. ~$533M liquidity against a multi-GW buildout costing billions does not close even with the convert.
• Froth signature deflating. Coverage skewed to penny-momentum desks (TimothySykes, StocksToTrade); an 8x move off $0.70 into a 52-wk high now reversing is peak-sentiment unwinding. Analyst targets cluster low: avg ~$5.25, range $3–$8 (Chardan $5.50 on 2026-05-27, HC Wainwright $5.50) spot ~$5.13 sits at the average, no margin of safety.

Setup & Price Structure

The clean breakout structure broke this week. Sequence: low-$4s launch base (~$4.00–$4.20) → $6.60 52-wk high → $6.10 close 2026-06-01 (+7.39%) → $5.93 on 06-04 → roughly $5.13 on 06-05 on the convert news, ~16% off the high. Market cap ~$3.58B puts P/S near 16x trailing with no profit path. The name is now mean-reverting toward its moving averages rather than extending above them, and the prior post-breakout shelf at $5.00 is the line in the sand it is being actively tested. This is a falling-knife tape on a dilution event with convert-arb short pressure layered on; the ACCELERATING + cluster-confirmed condition that would justify chasing strength is absent here. Below $5.00 the next reference is the $4.00–$4.20 launch base. A constructive re-set would need a higher low above $5.00 and a reclaim of the broken shelf on volume, ideally paired with a lease catalyst not a knife-catch into the financing overhang.

Catalyst Calendar (next 30 days)

• 2026-06-09 (est.): $400M convertible offering scheduled to close overhang event; convert-arb hedging established, watch whether the sell-the-financing pressure exhausts or extends.
• ~2026-06-18 (est.): 13-day greenshoe window for the additional $58M of notes lapses incremental dilution risk until then.
• Undated, "any week": first signed HPC/AI tenant lease (Panther Creek 350 MW / Sharon 110 MW / Moses Lake) the binary; a signing flips conviction up, continued silence compresses the multiple.
• Ongoing: Bitcoin price sensitivity KEEL still carries BTC on the balance sheet, so a BTC drawdown is an independent down-catalyst.
• No earnings in window: Q1 2026 printed 2026-05-11; Q2 2026 expected ~early-to-mid Aug 2026 outside 30 days, not a near-term blocker.

What Would Change Our Mind

A signed HPC/AI lease at any of the three sites is the upgrade trigger it converts "pipeline" into contracted, utility-like cash flow and would justify re-rating above the Street's $5.25 average toward the $8 high mark; that flips the read from cracking to re-accelerating. On the downside, a weekly close below $5.00 confirms the breakout shelf is lost and the financing selloff has become a trend break, pointing to the $4.00–$4.20 launch base. The thesis is also broken if Q2 (~Aug 2026) prints with still-zero signed leases, or if an at-market follow-on equity raise lands on top of the convert that would signal the $533M liquidity plus $400M notes still cannot fund the buildout and dilution is being forced at depressed levels. A BTC breakdown adds independent downside given the retained coins.

Correlation Notes

Trades as a high-beta member of the ex-miner AI-datacenter cohort IREN, WULF, CIFR, CORZ, APLD so peer lease prints (Applied Digital's $7.5B deal) and cohort drawdowns both spill over. Secondary correlation to Bitcoin via retained balance-sheet coins, making it a hybrid power-infra / crypto-beta name. Rates-sensitive on two counts: long-duration datacenter capex economics, and the new converts whose value (and arb hedging behavior) tracks rate/vol moves. Near-term, single-name price action is partly mechanical convert-arb desks shorting equity to hedge the $400M notes which can decouple the stock from the cohort until the offering closes ~2026-06-09. Broad risk-off / Nasdaq beta amplifies moves given the penny-origin float.

Current Thesis

(See header — narrative leg is the BTC-miner→AI-datacenter pivot; the financing has confirmed the funding gap while the lease binary stays unwon, flipping the tape to distribution.)