Figure AI / Unitree source-to-slide claim control — 2026-06-17

Date: 2026-06-17 Owner: Hugo / Genius Team Agent: Finance / Charlie AGT-002 Status: public-safe research artifact; slide/site claim-control layer Visibility: PUBLIC Primary use: morning review, slide 11 Tesla / Figure / Unitree comparison, /robotics/companies evidence cards Related files:

• knowledge/robotics-research-harness.md
• knowledge/robotics-sharing-backlog.md
• knowledge/robotics-codex-handoff.md
• knowledge/robotics-figure-unitree-evidence-v1.md
• knowledge/robotics-public-claim-bank-v1.md
• knowledge/robotics-oem-public-evidence-matrix-v1.md
• knowledge/robotics-morning-review-next-signal-matrix-2026-06-15.md

Public-safety: yes. No Hugo private portfolio data. No buy / sell / hold language. No external publishing action. Source discipline: 🟢 primary / 🟡 secondary / 🟠 estimate / 🔴 guess.

0. One-line answer

截至 2026-06-17，Figure AI 和 Unitree / 宇树不应该被包装成“谁赢”的横向排名，而应该被包装成两条不同的 S3/S4 证据曲线：Figure 的强项是 named-customer deployment KPI + manufacturing KPI；Unitree 的强项是公开低价硬件 + developer/data stack access。二者都还没有公开披露 S5 scaled-commercial-economics proof：contract value、repeat order、customer ROI/payback、uptime/intervention、gross margin、support burden。🟢 existing Figure / Unitree primary-source artifacts reviewed through 2026-06-17; 🟠 Charlie synthesis.

Approved public sentence:

Figure is the stronger deployment-evidence curve; Unitree is the stronger cost-access curve. Both are useful, neither is yet full commercialization proof.

Do not say:

Figure 已经证明 humanoid 大规模商业化；Unitree 低价已经证明 humanoid 会普及。

Reason: reviewed public evidence supports measurable S3/S4 progress, but not the full economic chain. 🟢/🟠 as of 2026-06-17.

1. Claim-control table for slide/site packaging

2. Slide-ready compression

Suggested title:

Figure vs Unitree: deployment proof vs cost-curve proof

Suggested subtitle:

Two useful evidence curves; neither yet closes the S5 economics gap.

Three-card layout:

1. Figure — deployment KPI

   • Proven: BMW 11-month deployment; 10-hour weekday shifts; 90,000+ parts; 1,250+ runtime hours; 30,000+ X3 vehicles contributed. 🟢 Figure official post, 2025-11-19.
   • Also proven: BotQ 350+ Figure 03 delivered; 1/day to 1/hour cadence; >80% EOL FPY; 99.3% battery-line FPY; 9,000+ actuators. 🟢 Figure official post, 2026-04-29.
   • Missing: robot count by customer, contract value, ROI/payback, intervention rate, repeat orders, revenue/margin. 🟢/🟠.

2. Unitree — cost-access curve

   • Proven: R1 AIR from US$4,900 / R1 from US$5,900; G1 from US$13.5K; H2 US$29,900. 🟢 Unitree official pages accessed 2026-06-10.
   • Also proven: H2 Plus / G1-D expose NVIDIA / Isaac / data-model workflow surfaces. 🟢 official pages.
   • Missing: shipments by product line, industrial uptime, customer mix, gross margin, warranty/support burden, repeat orders. 🟢/🟠.

3. Correct framing

   • Figure answers: “Can humanoids do measurable work at a named customer site?” S4, not S5.
   • Unitree answers: “Can humanoid bodies become cheap and accessible enough to widen experimentation?” S3/S4, not S5.
   • The comparison is evidence-type, not winner ranking. 🟠 synthesis.

Required footer:

Evidence map only. No winner ranking. No trade recommendation. Deployment KPI and low price are not the same as scaled-commercial-economics proof.

3. Signal vs noise

Signal

Noise unless upgraded

4. S5 evidence-debt checklist

Figure upgrades toward S5 only if public sources disclose at least 3 of the following together:

1. Customer-confirmed robot count, paid contract value, or revenue/backlog by deployment. 🟢 needed.
2. Repeat orders or multi-site expansion with same KPI package. 🟢 needed.
3. ROI/payback or labor/productivity savings confirmed by customer, not only vendor. 🟢/🟡 needed.
4. Uptime, failure distribution, intervention rate, safety incidents, and service burden over months. 🟢/🟡 needed.
5. Link from BotQ output to customer accepted units, revenue, gross margin, and field reliability. 🟢 needed.

Unitree upgrades toward S5 only if public sources disclose at least 3 of the following together:

1. Humanoid shipment volume by product line and customer type. 🟢 needed.
2. Gross margin and warranty/support cost by product category, ideally filing-grade. 🟢 needed.
3. Industrial customer deployment KPI: uptime, intervention, task output, repeat orders. 🟢/🟡 needed.
4. Paid software/data/tooling revenue or developer ecosystem traction, not only hardware sales. 🟢/🟠 needed.
5. Product quality / reliability evidence across non-demo environments. 🟢/🟡 needed.

5. What would change the morning-review answer

Upgrade the comparison if:

• Figure adds a second or third named customer with BMW-like runtime/task KPI plus robot count, repeat order and ROI/payback. 🟢
• Catalyst discloses material deployment count/value or customer-confirmed productivity data. 🟢/🟡
• Unitree files or publishes shipment / revenue / margin / support-cost data that show low-cost humanoids can scale without destroying economics. 🟢
• Unitree or a third-party customer discloses industrial deployment KPI comparable to Figure’s BMW metrics. 🟢/🟡

Downgrade the comparison if:

• Figure’s BotQ output accumulates without customer utilization or accepted-unit economics. 🟠
• BMW remains a one-off showcase and Catalyst remains terms-undisclosed. 🟠
• Unitree’s low price expands demos but not reliable paid deployments. 🟠
• Support cost, warranty cost, field failure or low gross margin overwhelms low-price adoption. 🟢/🟡

6. Common misconceptions

1. “Figure has BMW metrics, so humanoid economics are solved.”

   • Correction: BMW metrics are strong S4 deployment evidence, but not ROI/payback, contract value, repeat order, margin or intervention proof. 🟢/🟠

2. “Unitree is cheap, so Unitree captures the most value.”

   • Correction: low price can expand adoption and compress the industry cost curve, but value may migrate to software, data, compute, integration, deployment services, or end customers. 🟢/🟠

3. “Manufacturing KPI equals commercial demand.”

   • Correction: 350+ robots delivered and improved production cadence matter, but demand quality still requires customer accepted units, utilization, revenue and margin. 🟢/🟠

4. “Product specs decide the race.”

   • Correction: specs are lower-signal than task success, runtime, intervention, support cost, payback, repeat order and margin. 🟠 framework.

7. Think Deeper questions

1. If Figure continues to disclose deployment KPI but not contract economics, what minimum data should be enough to call the deployment repeatable?
2. If Unitree reduces humanoid access cost by an order of magnitude versus older six-figure lab robots, who captures the value: hardware OEM, compute stack, model/data layer, integrator, or customer?
3. Does cheap hardware accelerate data collection enough to improve autonomy, or does it mostly create demo proliferation?
4. Which signal comes first in a real return cycle: customer KPI, low hardware price, manufacturing yield, supplier revenue, or gross margin?
5. If Figure and Unitree converge, will the winning path look like “industrial deployment first, lower cost later” or “low-cost platform first, deployment proof later”?

8. Codex packaging guidance

Use this as a claim-control sheet, not as final prose.

Best use:

• Slide 11: Tesla / Figure / Unitree comparison card.
• /robotics/companies: Figure and Unitree evidence cards.
• Morning review speaker notes: prevent overclaiming Figure BMW or Unitree price points.

Do not use this to:

• Rank Figure vs Unitree as winners.
• Recommend trades or imply private-company investment action.
• Publish Hugo portfolio context.
• Treat vendor claims as customer-confirmed economics.

9. Source list

Primary sources / primary-source-derived artifacts:

• Figure, “F.02 Contributed to the Production of 30,000 Cars at BMW,” 2025-11-19: https://www.figure.ai/news/production-at-bmw 🟢; metrics captured in robotics-figure-unitree-evidence-v1.md; reviewed through 2026-06-17.
• Figure, “Ramping Figure 03 Production,” 2026-04-29: https://www.figure.ai/news/ramping-figure-03-production 🟢; metrics captured in prior artifacts; reviewed through 2026-06-17.
• Figure, “Figure Signs Agreement with Catalyst Brands to Scale Humanoid Operations,” 2026-05-26: https://www.figure.ai/news/figure-signs-agreement-with-catalyst-brands 🟢 for announcement; 🟠 for economics because terms undisclosed.
• Figure, “Introducing Helix 02: Full-Body Autonomy,” 2026-01-27: https://www.figure.ai/news/helix-02 🟢 for company claim; 🟠 for commercial implication.
• Unitree R1 product page: https://www.unitree.com/R1 🟢 accessed 2026-06-10 in prior artifacts; used for R1 AIR / R1 price anchors.
• Unitree G1 product page: https://www.unitree.com/g1/ 🟢 accessed 2026-06-10 in prior artifacts; used for G1 price / developer platform anchor.
• Unitree H2 product page: https://www.unitree.com/H2 🟢 accessed 2026-06-10 in prior artifacts; used for H2 price anchor.
• Unitree H2 Plus product page: https://www.unitree.com/H2plus 🟢 accessed 2026-06-10 in prior artifacts; used for Jetson / Isaac / TeleOp / Sim workflow claims.
• Unitree G1-D product page: https://www.unitree.com/G1-D 🟢 accessed 2026-06-10 in prior artifacts; used for data acquisition / model training / inference workflow.
• Charlie synthesis across existing public-safe artifacts, 2026-06-17. 🟠.

10. Public-safe flag

Public-safe: yes, as an evidence map and claim-control sheet.

Exclude:

• Hugo private portfolio data, position weights, entry logic, watchlist sizing or private rationale.
• Buy / sell / hold language for public assets or private-company investment solicitations.
• Supplier rumors, private channel checks, paid-report excerpts, or unverified customer claims.
• Claims that Figure has proven scaled commercialization or Unitree has proven low-cost deployment economics.
• Claims that vendor-published autonomy demos are customer-confirmed uptime/intervention proof.