Ekso Bionics

Phase A — Understand the business (the historical Ekso Bionics medical-robotics company)

Company Overview

Ekso Bionics Holdings was a Richmond/San Rafael, California robotic-exoskeleton company (founded 2005 as Berkeley Bionics; public via reverse merger 2014) that designed, manufactured and serviced wearable powered exoskeletons. It ran two segments:

• EksoHealth — medical/rehabilitation exoskeletons. The flagship EksoNR (successor to Ekso GT, FDA-cleared 2016 for stroke & spinal-cord injury; extended to acquired brain injury in 2020 and multiple sclerosis in 2022 — the first exoskeleton cleared for MS rehab) is a clinic-based gait-training robot. In Dec 2022 Ekso bought the Indego lower-limb exoskeleton line (plus the Human Motion & Control unit) from Parker Hannifin for $10M, adding a lighter personal/home device and a path to individual-patient (Personal Health) sales.
• EksoWorks — industrial exoskeletons (e.g. EVO) letting able-bodied workers perform overhead/repetitive labor with less fatigue. Always the smaller, slower leg.

How it made money: capital sales and leases of devices to rehabilitation hospitals (the "Enterprise Health" channel — ~75–80% of near-term revenue), plus service contracts, extended warranties, and clinical-training/CEU programs; and, increasingly, personal-device sales to individuals funded by insurance/Medicare reimbursement ("Personal Health"). In late 2025 it named National Seating & Mobility (NSM) the exclusive U.S. Indego Personal distributor in the complex-rehab-technology channel. Customers = rehab hospitals, VA facilities, universities (e.g. Vanderbilt, Burke/Weill Cornell), and individual SCI/stroke/MS patients. Contract structure was transactional and lumpy — big-ticket ($70k–$150k/unit class) capital equipment with no meaningful recurring-revenue base, which is the structural root of its quarter-to-quarter volatility.

Bottom line on the model: a real, FDA-cleared, clinically-validated product with a genuine unmet-need story — trapped in a business that was too small and too lumpy to ever cover its own cost structure. Peak annual revenue of $18.3M (FY2023) against tens of millions in cumulative losses is the whole tragedy in two numbers.

Supply Chain

Named-stakeholder map (mostly, as no supply-chain filing was ingested):

• Upstream inputs: actuators/motors, harmonic-drive gearing, Li-ion battery packs, structural aluminum/composites, motor controllers, sensors/IMUs, embedded compute. Ekso was a contract-manufactured / assembled hardware company — it designed and integrated; it did not fab its own silicon or motors. Specific component suppliers are not publicly disclosed at the SKU level.
• The company: design, systems integration, firmware, clinical/regulatory (FDA 510(k)), final assembly and QA in California.
• Channel / distribution: direct sales force into rehab hospitals (Enterprise Health); NSM as exclusive U.S. Indego Personal distributor (Personal Health, from 2025); international distributors across EMEA/APAC.
• End customers: rehabilitation hospitals & clinics, the VA, academic medical centers (Vanderbilt, Burke/Weill Cornell partnerships), and individual patients (SCI, stroke, ABI, MS) whose purchases hinge on third-party reimbursement (Medicare/commercial insurance — Ekso secured Medicare coverage for the Indego Personal device around Q4-2024).

Chokepoints / single-source dependencies: (1) Reimbursement is the true bottleneck — the personal-device market only exists to the extent CMS/insurers pay; a coverage-policy change is an existential input, not a cost line. (2) FDA clearance gates every indication and every product iteration. (3) Concentrated institutional buyers with long, budget-cycle-driven purchasing = revenue you cannot forecast. This lens is deliberately specific: the exoskeleton BOM is commoditized; the scarce inputs are regulatory clearances and payer coverage, and Ekso controlled neither.

Competitive Advantages (moats)

What Ekso actually had:

• First-mover clinical evidence & installed base. Ekso GT was the first exoskeleton FDA-cleared for stroke + SCI (2016); EksoNR was the first cleared for MS (2022). Years of peer-reviewed clinical data and hundreds of clinic placements created real switching friction inside rehab departments (staff training, protocols, workflow) and a ~17% share claim in medical exoskeletons.
• Regulatory estate — a stack of 510(k) clearances across indications that a new entrant must replicate.
• Brand among rehab clinicians (the "Ekso" name is near-generic in gait-training circles).

Why the moat was shallow in practice:

• No scale economics. At <$20M revenue the fixed cost of R&D, regulatory, and a direct sales force could never be amortized — the classic sub-scale-medtech trap. A moat that cannot pay for itself is a liability.
• Weak pricing power / bargaining position. Ekso needed the hospitals (concentrated, budget-constrained buyers) far more than they needed Ekso; it needed CMS coverage more than CMS needed the device. Bargaining power over both suppliers (commodity components) and customers (institutional, price-sensitive) was structurally weak.
• Contested category. Direct rivals ReWalk/Lifeward, Cyberdyne (HAL, bioelectric), Ottobock, and DIH Medical all fight the same clinics; personal-device reimbursement wins accrue to whoever gets the CMS nod, not to an incumbent by right.

Verdict: a narrow, non-durable moat. Real IP and evidence, zero cost advantage, no network effect, no lock-in strong enough to outrun the cash burnCash burnHow much more cash goes out than comes in, per period. The clock on a company with no profits.. Moats that don't compound into unit economics are just nice press releases — and this one ended in a reverse merger.

Segments

No our figures data was ingested (empty scaffold), so segment splits are and partial. Ekso reported EksoHealth vs EksoWorks in filings and, operationally, framed revenue as Enterprise Health vs Personal Health:

• EksoHealth / Enterprise Health = the overwhelming majority — management guided ~75–80% of near-term revenue to Enterprise Health. This is clinic device sales (EksoNR) + service.
• Personal Health = the growth bet — Indego Personal into the home/community channel via NSM, unlocked by 2024 Medicare coverage. Small base, higher strategic hope.
• EksoWorks (industrial) = the persistently sub-scale segment; never the story.

Trend & cause: total revenue decelerated then rolled over — FY2022 $12.9M → FY2023 $18.3M (+42%) → FY2024 $17.9M (−2%) → 2025 collapsed quarterly (see Lens 5). The FY2023 jump was partly the Indego (Parker Hannifin) acquisition broadening the line; FY2024's flat-to-down print showed the acquisition did not create durable growth. Geographic split was not cleanly sourced; U.S. was the primary market with international distribution secondary. Hard-requirement honesty: precise per-segment dollar splits are n/a (would require ingesting the 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes. segment footnote).

---

Phase B — Measure performance (what the numbers and the tape said)

Earnings Result (the trajectory, then the final prints)

All figures from press releases (GlobeNewswire / StockTitan / Nasdaq / IR). The arc is a company shrinking into distress:

.

What the prints say:

• Revenue never scaled and then reversed. The "record" quarters ($5.0M Q2-24, $5.1M Q4-24) were records against a tiny base; 2025 saw revenue halve in Q2 to $2.1M, bounce to $4.2M in Q3, then fall again to $2.14M in Q1-2026. Extreme lumpiness = no recurring revenue.
• Margins were the one genuine bright spot — gross margin climbed from ~49% (2023) to ~53% (2024) to 60.3% (Q3-2025), driven by "high-margin Enterprise Health" mix and lower device cost/service improvement. But 60% gross margin on $4M of quarterly revenue cannot cover the opex of an FDA-regulated hardware company.
• Persistent losses & cash burn. Net losses every year (−$15.2M FY23, −$11.3M FY24); cash usage $9.8M (2024) improving from $12.1M (2023). Cash was $5.2M at Q2-2025 — a runwayRunwayHow long the cash lasts at the current rate of spending. It shortens the moment spending rises, which is why a figure taken from a quiet quarter flatters. measured in a couple of quarters.
• Balance-sheet distress → dilution machine. Repeated raises: warrant exercises (+$3.8M Q1-25), a $3.7M registered direct at $4.81/sh (Oct 2025), a $5.852M private placement (Jan 2026). Then the 1-for-15 reverse split (effective Jun 2, 2025) to cure a Nasdaq minimum-bid deficiency (notice received Dec 12, 2024), collapsing share count from ~35.5M to ~2.4M.
• Market reaction / what was priced in: the equity was already left for dead — stock hit $0.2309 (May 22, 2025) pre-split; the reverse split and the eventual APLD reverse-merger were the market pricing terminal value of the operating business at ~zero and re-rating the shell on the datacenter deal (stock to ~$10.50 by Mar 2026 on merger optics — see Lens 8). Unusual vs history: the entire 2025→2026 move is corporate-action-driven, not fundamentals-driven.

Earnings Calls (sentiment trend)

No transcripts ingested; sentiment is from call recaps (Investing.com transcripts, Yahoo highlights). Tone arc across the last several calls:

• 2023–early 2024 ("record revenue, aims for growth"): management optimistic — Indego integration, MS clearance, "continuum of care" narrative, record annual revenue.
• Late 2024 (Q3/Q4): "mixed" results, record Q4 but full-year flat; pivot language toward Personal Health / Medicare coverage and cost discipline.
• 2025 (Q2/Q3): revenue drop yet stock rose on cost/margin improvement and the AI angle — the story shifted from "sell more exoskeletons" to "foundation model for human motion" (NVIDIA Connect, May 2025) and to strategic alternatives. The recurring new phrase was AI / foundation model; the thing they effectively stopped promising was organic revenue growth.
• 2026: the narrative is fully corporate-action — merger with Applied Digital, name change, strategic separation of the exoskeleton unit.

Sentiment read: a classic distressed-microcap tonal drift — from "growth" → "mixed, disciplined" → "we have an AI story" → "we're selling the shell." Management leaned on narrative substitution (AI foundation model) as fundamentals deteriorated.

Comps

Peer set = pure-play medical/powered-exoskeleton names (the only honest comparables; the successor's true comps are datacenter/neocloud — see the overlay note). Multiples are or n/a; no multiple is fabricated.

Read: the entire listed pure-play exoskeleton cohort is sub-scale and unprofitable — Lifeward, the "largest," carries a ~$10M market cap on ~$24M revenue and negative earnings. This is a category, not just a company, problem: a decade in, no Western pure-play has reached self-funding scale. Ekso screened as one of the healthier-margin names but smallest-revenue — and still had to sell its listing. Multiples deliberately left n/a: with negative EBIT/EPS across the cohort, P/E and EV/EBIT are meaningless, and I did not source clean EV/Sales figures this run.

Stock-Price Catalysts (moves >5% over the last ~5 years)

Mostly:

• 2014 all-time high $863.10 (Feb 24, 2014, pre-all-splits) → a ~decade-long, near-total decline. The dominant "catalyst" was chronic dilution and cash burn, not events.
• Dec 2022 — Indego/Parker Hannifin acquisition ($10M): strategic pop, expanded product line.
• 2022 — EksoNR MS clearance: positive clinical catalyst.
• Dec 2024 — Nasdaq minimum-bid deficiency notice → Jun 2025 1:15 reverse split: the classic distress sequence; stock hit $0.2309 (May 22, 2025).
• 2024 — Medicare coverage for Indego Personal: genuine positive (reimbursement is the category's oxygen), but too small to change the trajectory.
• May 15, 2025 — NVIDIA Connect acceptance / "foundation model for human motion": the AI-narrative catalyst; stock re-rated on story despite falling revenue.
• Dec 29, 2025 (LOI) → Feb 15, 2026 (definitive) → May 5, 2026 (close) — Applied Digital reverse merger: the decisive re-rating. Stock ran from ~$1.19 (end-2024) to a 52-wk high $13.50 (Mar 11, 2026) on merger/AI-datacenter optics. H.C. Wainwright downgraded to Neutral on the merger.

What the tape reveals about this name: for its final two years the market stopped reacting to exoskeleton fundamentals entirely and traded on (a) survival/dilution mechanics and (b) the AI-datacenter reverse-merger. The operating business became irrelevant to the share price well before it was formally sidelined — the equity was a corporate-action instrument, not a medtech.

---

Phase C — Judge people & books

Management (historical Ekso)

:

• Steven Sherman (Chairman from 2014; Exec Chair; briefly CEO Jan–Dec 2022, age 76 in 2022) — a long-tenured financier who invested significant personal capital since 2013 (real skin in the game). His long chairmanship spanned the entire multi-year value destruction from the 2014 peak, so alignment did not translate into shareholder returns.
• Scott Davis (CEO Dec 2022 → through the merger; prior President/COO from Jan 2022; EVP Strategy & Corp Dev from Apr 2021; background: Sierra Wireless enterprise sales, SGD Executive Services consulting) — a strategy/corp-dev & sales operator, not a clinical/robotics founder. His mandate read increasingly as manage decline and find a strategic exit — which is precisely what the AI-narrative pivot and the Applied Digital reverse merger delivered.

Assessment against the five tests:

1. Track record: managed a shrinking sub-scale business; the "achievement" was cost discipline + margin improvement + engineering an exit, not building a growth company.
2. Tenure & skin in the game: high (Sherman's personal capital; long tenures) — but alignment without results.
3. Capital allocation: the $10M Indego acquisition (2022) did not produce durable growth; serial dilution (warrants, RDOs, private placements) at ever-lower prices destroyed per-share value; the reverse split was defensive. ROE/ROIC were persistently negative. On balance, value-destructive by necessity of the business, though the final reverse-merger arguably salvaged residual shell value for stranded holders.
4. Red flags: heavy, repeated dilution; narrative substitution (leaning on the NVIDIA/AI story as fundamentals fell); an eventual reverse merger with a related-adjacent AI-datacenter vehicle where legacy holders end up with ~3% of the new entity.
5. Archetype: professional managers / financiers, not founder-operators — appropriate for a wind-down / strategic-exit stage, which is exactly the stage this was.

Successor note: ChronoScale is run by Cenly Chen (CEO from 2026-05-06), an Applied Digital-side appointment — confirming the shell is now operated for the datacenter business, not the exoskeleton team.

Forensic Red Flags

No filings on the shelf, so income-statement/balance-sheet forensics are -level and directional, not line-item audited:

• Cash-flow vs earnings: losses were cash losses (operating cash usage ~$9.8M FY24) — not an accruals-masking situation; the problem was real cash burn, not aggressive revenue recognition.
• Revenue quality: lumpy capital-equipment sales; the Q2-2025 collapse to $2.1M then Q3 bounce to $4.2M shows revenue recognition tied to device shipment timing — volatile but not obviously manipulated. Watch item historically: service vs device mix and deferred-revenue treatment on multi-year service/warranty contracts (n/a — not verified without the 10-K footnotes).
• Dilution / SBC: the dominant "red flag" is relentless share issuance (warrants, RDOs, private placements, reverse split). This is dilution risk, not accounting fraud — but for equity holders the effect (near-total value transfer away from legacy shares) is the same.
• Going concern: a sub-$20M-revenue company burning ~$10M/yr with ~$5M cash is a textbook going-concern watch; the reverse merger was, in effect, the going-concern resolution.

Regulatory findings (required sub-section):

• SEC (EDGAR LR/AAER): the research-layer file regulatory/regulatory-findings.md records total_sec_findings: 0 and notes Ekso has cik: null in the index so no EDGAR enforcement search was run by the script. (Historically Ekso did file under CIK 0001549084; a future refresh could confirm directly, but no enforcement action is known.)
• Non-SEC (FTC/DOJ/FDA/CFPB): web search "Ekso Bionics" (FTC OR DOJ OR FDA OR settlement OR fine OR penalty) enforcement surfaced no material enforcement action, consent decree, or penalty. FDA interactions are product clearances (510(k)), not adverse actions — i.e. regulatory approvals, the opposite of enforcement.
• 10-K Item 3 (Legal Proceedings): not ingested to the shelf this run (n/a — 10-K not on shelf); the most-flagged "risk" in the FY2024 10-K per coverage was the merger/strategic-transaction risk itself, not litigation.
• Conclusion: No material regulatory or legal findings — verified via the research-layer regulatory file (SEC LR/AAER = 0, no CIK) and web search across FTC/DOJ/FDA as of 2026-07-06. The dominant investor risk was never enforcement; it was dilution and going-concern.

---

Phase D — Project & stress-test

Forward Projection (successor-overlay framing — the honest version)

A forward EPS projection for "Ekso Bionics" is not meaningful — the entity is gone. The legacy exoskeleton business is being divested from ChronoScale and, on a standalone basis, had no realistic path to positive EPS (sub-$20M revenue, structural losses). So this lens splits:

(a) The historical operating business — terminal, no forecast. Standalone Ekso EPS trajectory: negative, indefinitely. FY2024 −$0.56; 2025 quarterly EPS between −$1.24 and −$1.93 (post-split share counts). Base/bull/bear all converge on "does not reach profitability before running out of cash" — which is why it was sold.. No our model create — there is nothing scoreable here (and the mandate forbids it in watchlist mode).

(b) What a CHRN holder actually owns now — an Applied Digital datacenter proxy. Post-merger, "EKSO/CHRN" value is ~97% driven by Applied Digital's cloud/GPU-compute business, not exoskeletons. That business is a CoreWeave-anchored hyperscale-datacenter operator: at Ellendale, ND (Polaris Forge 1), APLD signed 250MW + 150MW = 400MW of CoreWeave leases with ~$11B total contracted lease revenue over ~15 years, financed partly by a $1.59B 7.0% senior secured notes (due 2031). So the CHRN thesis is a leveraged, single-tenant-concentrated (CoreWeave) AI-datacenter build-out — an entirely different risk profile (capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. intensity, tenant concentration, financing/rate risk, execution timing on 2025–2027 ready-for-service dates) than a medtech.

Projection verdict: value the successor as a neocloud/datacenter proxy with heavy CoreWeave counterparty concentration and a levered balance sheet, not as an exoskeleton company. The legacy business is a rounding error / divestiture candidate. — a proper CHRN forecast belongs in a dedicated Applied Digital / ChronoScale deep-dive (recommended below), not here.

Bull vs Bear

Bull case (for the historical exoskeleton business — now moot): real FDA-cleared products, first-mover clinical evidence, ~17% share, improving gross margin (60%+), a growing personal/Medicare-reimbursed channel, and an interesting AI foundation-model-for-human-motion optionality (NVIDIA Connect). In a world where the category inflects (the medical-exoskeleton TAM is quoted at ~$1.4B in 2025 → multi-billion by mid-2030s at ~17–19% CAGR ), a scaled leader could matter.

Bear case (which won): three permanent-impairment risks all fired — (1) structural sub-scale: revenue never crossed the threshold to cover a regulated-hardware cost base; (2) reimbursement/adoption gating: the personal market depends on payers, and adoption stayed too slow; (3) capital-markets dependence: chronic dilution + a going-concern balance sheet meant the equity was perpetually raising at lower prices, ending in a 1:15 reverse split and near-zero price. The market baked in terminal value ≈ 0 for the operating business.

Pre-mortem (it already happened): 18 months out, the thesis "Ekso scales its exoskeleton franchise" broke because the business was too small to self-fund and payer/adoption ramps were too slow; management, unable to grow into the cost base, substituted an AI narrative and then sold the listing to an AI-datacenter roll-up (Applied Digital), leaving legacy holders with ~3% of a completely different company.

Multiples too high? For the operating business, any positive equity value was too high relative to a zero-terminal-value standalone — which is why it was effectively given away. For the successor (CHRN), the risk inverts: you're now paying datacenter-buildout multiples on a levered, single-tenant-concentrated entity, and the exoskeleton legacy is noise.

Contrarian view (what the market refused to see, and then saw all at once): the market spent years pricing EKSO as a struggling-but-going medtech; the actual value event was a shell/listing transaction. The lesson: for terminal-stage microcaps, the equity can become a reverse-merger option on someone else's business — which is exactly what happened, and the ~$0.23 → ~$13.50 round-trip proves the operating fundamentals were never what mattered at the end.

Devil's Advocate (short-seller)

Dismantling any residual "own EKSO/CHRN for the exoskeleton upside" bull case:

• The exoskeleton business is being divested — so buying CHRN for exoskeleton exposure is buying the thing management is actively removing.
• Revenue concentration / counterparty risk migrated, it didn't disappear. Old Ekso's risk was concentrated institutional buyers; new CHRN's risk is concentrated in one tenant, CoreWeave, for the bulk of that $11B contracted revenue — a single-counterparty AI-datacenter bet. If CoreWeave's own economics or AI-capex cycle wobble, the whole lease book is exposed.
• The moat was always weak (Lens 3) — commodity BOM, payer-gated demand, no cost advantage; bulls who believed in an exoskeleton moat were underestimating Lifeward, Cyberdyne, Ottobock and the reimbursement gatekeepers.
• Worst capital-allocation pattern: serial dilution at falling prices + a reverse split + a reverse merger that hands ~97% to the incoming party. Legacy equity was structurally set up to be diluted to a sliver.
• What must hold for today's (CHRN) price: flawless 400MW datacenter execution on schedule (2025–2027), CoreWeave's continued creditworthiness, and that a $1.59B 7% notes balance sheet stays serviceable through the AI-capex cycle. If AI-datacenter demand disappoints by 20–30%, a levered, single-tenant operator is exactly where the pain concentrates.
• Single scenario that permanently impairs: a CoreWeave stumble (or an AI-capex air-pocket) hitting a highly-levered ChronoScale before the lease revenue fully ramps. Plausibility: material — this is the central risk of the entire neocloud/datacenter-leasing complex, and CHRN is a concentrated expression of it.

Management Questions (15, ordered by information value)

Directed at ChronoScale / Applied Digital leadership (Cenly Chen; APLD), since that is who now controls the listing — the legacy Ekso team is exiting:

1. What is the definitive plan and timeline for the legacy Ekso exoskeleton business — outright sale, spin, or wind-down — and what proceeds (if any) accrue to CHRN shareholders?
2. What share of ChronoScale's contracted $11B lease revenue is CoreWeave, and what is your single-tenant concentration mitigation?
3. What is CoreWeave's contractual credit protection (deposits, parent guarantees, take-or-pay floors) if it defers or defaults?
4. Give the ready-for-service schedule for each tranche (100MW Q4-2025, 150MW mid-2026, 150MW 2027) and the penalty/adjustment terms for slippage.
5. What is pro-forma net leverage post the $1.59B 7% 2031 notes, and what covenants could bind if lease ramp slips?
6. What is the power-cost and power-availability position at Ellendale (Polaris Forge 1), and how is it hedged?
7. What incremental capex (self-funded vs financed) is required to bring the full 400MW online, and what's the funding plan?
8. Why keep a ~3% legacy float at all — what is the plan for the minority holders inherited from Ekso?
9. What is the customer-diversification roadmap beyond CoreWeave, and the realistic timeline?
10. How does ChronoScale differentiate versus CoreWeave itself, Nebius, Crusoe, and other neocloud/datacenter lessors competing for the same GPU-compute demand?
11. What are the contracted vs speculative portions of future capacity — how much is un-leased "build-it-and-they-will-come"?
12. What governance protections exist for public minority holders given Applied Digital's ~97% control?
13. Does the NVIDIA relationship (legacy NVIDIA Connect / any GPU-allocation priority) carry over and confer any real GPU-supply advantage?
14. What is the AI foundation-model-for-human-motion IP worth in a sale of the exoskeleton unit — is it bundled or retained?
15. What is the capital-return / dilution policy for CHRN over the next 24 months?

---