Written by Claude (Anthropic’s AI) from public sources · Willem owns DUOT · AI can make mistakes — numbers in brackets point to the sources · Not financial advice
| Tailwind | Why not in consensus | % chance (36M) | Potential impact |
|---|---|---|---|
| Axe/hyperscaler repeat orders >100 MW | Only contracted programs counted | 25% | +$8–15/share |
| Axe minority investment funds sites non-dilutively | Non-binding term sheets | 40% | +5–10% |
| Sector re-rating to 14x+ EV/EBITDA | Needham sector avg 10.8x | 20% | +$5–8/share |
| Headwind | Why not in consensus | % chance | Impact |
|---|---|---|---|
| Axe payment/ramp failure | Counterparty risk rarely modelled | 30% | −50% or worse |
| Equity raise at lower price | Capex need exceeds cash | 45% | −15–25% |
| Power/interconnect delays push revenue to 2028 | Guidance assumes 2026 deployment | 35% | −20–30% |
| Item | Value |
|---|---|
| Business now | Edge AI data-center colocation, modular 10–30 MW sites (Columbus, GA campus first); legacy rail (RIP portals) sold to Sandbank/Acosta 2026-08-05 (effective 2026-06-30), now "DuosTI" |
| Q2'26 revenue mix | Technology Solutions $3.23M (52%), Services & Consulting (related parties) $2.91M (47%), Hosting $0.03M |
| Geography | US only (Georgia, Texas planned) |
| Top customers | Axe Compute (GPU + hosting), unnamed investment-grade hyperscaler (10 MW), New APR Energy (declining related-party service revenue) |
| Moat | Weak/unproven. Type: speed to power/site + contracts. Evidence: contracted MW only; no proprietary tech after rail sale. Durability: contract term 5 years; edge colocation is competitive and capital-intensive. |
Interpretation: this is effectively a new company with an old ticker; legacy revenue history is of little use.
| Q2'25 | Q1'26 | Q2'26 | H1'26 | |
|---|---|---|---|---|
| Revenue | 4.77 | 2.14* | 6.18 (+30% y/y) | 8.32 (−4%) |
| Gross margin | 37.3% | n/f | 55.8% | 54% (4.50/8.32) |
| Operating income | −1.54 | −3.18* | +0.05 | −3.13 |
| Adj. EBITDA | not reported | n/f | 0.5 | n/f |
| Net income (total) | −3.52 | n/f | +47.84 (APR gain $53.2M) | +44.35 |
| Share count (period end) | n/f | n/f | 31.27M | 20.45M at Dec-25 |
| Cash | n/f | 33.0 | 112.3 | – |
| *Derived by subtracting Q2 from H1; Q1 originally reported as $2.72M including since-divested rail revenue, so not comparable. n/f = not found. FY25 revenue $28M (company release Feb 2026, as reported by Barchart). |
Other H1'26: operating cash flow +$11.4M; financing inflow $112.2M (stock $120.1M gross, $8.0M costs); deposits on equipment $68.8M, real-estate deposit $5.8M; PP&E $29.7M; total assets $241.3M; total liabilities $33.9M; equity $207.4M; income-tax payable $5.0M; long-term contract liabilities $18.8M. FCF margin and SBC % revenue: not found (SBC not extracted from cash-flow statement). ROIC: not meaningful (invested capital mostly fresh cash and deposits). Debt maturities: none material (notes payable $0.33M); the former $98.1M GPU facility was assumed/satisfied by Axe.
| Item | Result |
|---|---|
| Q1'26 revenue | $2.72M vs consensus ≈ $9.4–9.6M → big miss; EPS −$0.15 vs −$0.03 expected |
| Q2'26 | Revenue $6.18M; consensus not found |
| FY25 revenue guide | $28M delivered "in line with guidance" |
| FY26 guide >$50M | Reaffirmed after Q1 miss and Q2; H2-weighted |
| Tone | Shifted 2025→2026 from rail/AI-vision to GPUaaS (March) to colocation-only (October): three strategic frames in ~9 months |
| Recurring Q&A themes (from third-party summaries, transcript not obtained): timing of MW energisation, Axe terms, financing. Prior-year beats/misses beyond Q1: not found. |
| Sub-score | Weight | Score | Reasoning |
|---|---|---|---|
| Business quality & moat | 20% | 40 | Rail moat sold; colocation moat unproven, one-customer ramp |
| Growth | 20% | 80 | Contracted path from $28M (FY25) to >$50M (FY26) to ≥$160M (2027 framework) |
| Financial strength | 15% | 62 | $112M cash, ~no debt, but $69M deposits and large unfunded build-out |
| Valuation | 20% | 55 | EV ≈ $162M vs $611M contracted payments; but unproven delivery and dilution |
| Mgmt & capital allocation | 15% | 40 | Repeated pivots, Q1 miss, new CFO/COO, related-party dealings, +53% shares in 6 months |
| Sentiment & momentum | 10% | 50 | −44% from 52-wk high $15.28; range $6.17–15.28 |
| Weighted ≈ 55. Interpretation: growth is real on paper; execution and counterparty risk keep it "fair". |
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|
| Axe Compute concentration / counterparty | Medium-High | Very high | Axe reports $317M customer prepayments; GPU sale instalments $42.9M/60 months — but credit quality of Axe not verified |
| Guidance miss (H2 needs ~$42M) | High | High | Backlog ≈$28M Tech Solutions + $43.5M bookings quoted |
| Dilution / financing gap | High | High | $112M cash; Axe minority-investment term sheets (non-binding) |
| Construction/power delays | Medium | High | Modular design; deposits paid |
| Governance / related-party | Medium | Medium | Not disclosed in detail |
| Liquidity/volatility (small cap, ~$270M) | High | Medium | – |
| Tax (income-tax payable $5.0M) | Low | Low | Accrued |
| Bear | Base | Bull | |
|---|---|---|---|
| Operating MW 2029 | ~30 | ~100 | ~200 |
| Revenue 2029 | $60M | $200M | $350M |
| EBITDA margin | 33% | 45% | 48% |
| EBITDA | $20M | $90M | $170M |
| Exit EV/EBITDA | 8x | 11x (Needham sector avg 10.8x) | 14x |
| Net debt | $120M | $500M | $1,000M |
| Shares | 40M | 42M | 48M |
| EPS/FCF per share | negative | ≈ breakeven FCF (growth capex) | positive |
| Dividends | 0 | 0 | 0 |
| Price (2029) | $3.5 (incl. asset floor) | $11.7 | $28.8 |
| Probability | 35% | 50% | 15% |
| Probability-weighted price $11.4 → ≈ +33% (no dividends) → Upside Score 57 (band 30–60% → 56–75). | |||
| Reverse DCF (rough): EV ≈ $162M compounding at 12% for 3 years ≈ $228M; at 10x exit that implies only ≈ $23M of 2029 EBITDA (≈ 25% of my base case), assuming no further net debt. Interpretation: the market assumes much of the contracted build-out is not delivered or is funded dilutively. |
| Metric | Management | Consensus | Gap |
|---|---|---|---|
| FY26 revenue | >$50M (reaffirmed 2026-10-05) | not found | n/a |
| FY26 adj. EBITDA | positive each remaining quarter | not found | n/a |
| FY27 revenue | ≥$160M framework (formal guide with Q3) | not found | n/a |
| Price target | – | Ascendiant $14 (Dec-25); 2-analyst avg $25.5 (stockanalysis); MarketBeat "Hold", $24 | targets ≫ price |
| Credibility: Q1 missed badly; FY25 guide met; strategic story changed 3 times. Incentives and compensation: not found. Insiders: last open-market sale found was CFO Goldfarb May 2025 (~$72K); no 2026 open-market buys/sells found; a code "J" transaction ~261K shares on 2026-04-01 unexplained. Interpretation: absence of insider buying near $9.50 raise is neutral-to-slightly negative. |
| Peer | Data found |
|---|---|
| Applied Digital (APLD) | EV/EBITDA fwd ≈ 135x (May 2026, Trefis); debt $2.8B vs ~$11B market cap |
| WhiteFiber (WYFI, also in portfolio) | Mkt cap $670M, EV ≈ $985M (Sep 2026); fwd EV/EBITDA ≈ 5.8x (Apr 2026, Seeking Alpha) |
| Galaxy (GLXY) Helios | SOTP; Benchmark ≈ $21/share for contracted Helios |
| Sector | Needham: 10.8x 2028E EV/EBITDA; M&A references 20–30x (advisor marketing, low weight) |
| Growth/margin comparisons for peers: not found. Market share: not applicable at this scale. Substitutes: hyperscale campuses, neoclouds, on-prem. Entrants: bitcoin-miner conversions. |
No buybacks or dividends. Raised $120.1M gross in H1 (incl. $55M at $9.50, TD Cowen/Cantor, 2026-06-17); $8.0M issuance costs (6.6%). Sold APR stake ($50.4M) and rail business. Capex: $2.5M PP&E + $68.8M deposits + $5.8M real estate. M&A: Columbus facility acquisition pending. Dilution: shares +53%, diluted weighted 35.5M.
Institutions 25.6% (Simply Wall St) to 42.6% (MarketBeat) — sources conflict; insiders ~12% (Simply Wall St). Short interest: latest dated figure Aug 2025 (510K shares); current not found. Analysts: Ascendiant Buy $14; 2 analysts avg $25.5 (stockanalysis, 2026-10-07); another aggregator shows Hold. Momentum: $8.56 vs 52-wk $6.17–15.28; index comparison not found.
Power permitting and interconnection; export controls on Nvidia GPUs (indirect); tax payable $5.0M on APR gain; related-party disclosure; energy use/ESG scrutiny of AI data centers. Legal proceedings: not found.
~2026-11-11 Q3 results + formal 2027 guidance (est.); ~mid-Nov 0Lat exclusivity ends; Q4 2026 first 25 MW deployment; Columbus facility close; FY26 results ~Mar 2027; H2'26 DDC cooling deployment at Georgia campus.
| KPI | Current | Threshold |
|---|---|---|
| FY26 revenue guide | >$50M | break < $35M |
| Q3'26 revenue | – (Q2 $6.18M) | confirm ≥ $15M |
| MW energised | 0 confirmed (25 MW planned 2026) | ≥ 10 MW by Q1'27 |
| Cash | $112.3M (30 Jun) | < $50M without financing = red flag |
| Share count | 31.5M (+3.8M PFW) | > 45M by 2027 = red flag |
| Axe receipts | instalments $0.7M/mo | any miss = breaker |
| Gross margin | 55.8% | < 40% |
Weight 4.3% with Risk 74 — high risk per unit weight. Overlaps: WYFI (AI colocation), IREN, BTDR, SHAZ, KEEL, AIBZ, GRRR (AI infrastructure/GPU hosting) and AVGO/CRDO (AI capex) — correlated to AI-infrastructure sentiment and financing conditions.
The market is right to be sceptical of a story that changed three times, has 2 quarters of history, depends on one counterparty and needs ~8x H1 revenue in H2. It may be underpricing the contracted $600M+ of payments and the clean balance sheet (EV ≈ $162M). Score 55 reflects strong growth and cash, offset by moat, governance and counterparty weaknesses.
Confidence: Low-Medium. Not found: consensus revenue/EPS/FCF; SBC; FCF; 5-year valuation ranges; short interest; current insider and institutional detail; verbatim Q2 transcript (relied on 8-K prepared remarks + summaries); Axe Compute's financials/ownership ties; reason for 2026-10-07 drop; Q3'25–Q4'25 quarterly data. Source conflicts: backlog $25M vs $28M; 55 MW vs 65 MW (MarketBeat). I used 8-K figures where they differ.