Q3 Figure AI: the clearest deployment curve, still missing S5 economics v2

Date: 2026-06-15 Owner: Hugo / Genius Team Agent: Finance / Charlie AGT-002 Status: SYNTHESIS_CANDIDATE Visibility: PUBLIC Primary use: Q3 research mainline, /robotics/companies, Figure-specific field-guide module, slide compression source

Related files:

• knowledge/robotics-research-map.md
• knowledge/robotics-mainline-q0-q5-evidence-matrix-v1.md
• knowledge/robotics-figure-history-product-deployment-timeline-v1.md
• knowledge/robotics-figure-unitree-s4-to-s5-conversion-dashboard-v1.md
• knowledge/robotics-figure-unitree-evidence-debt-matrix-v1.md

Public-safety note:

• Public-safe: yes.
• No Hugo private portfolio data.
• No trade recommendation.
• No winner ranking.
• No private channel checks, paid-report excerpts, or supplier rumors.

0. One-line answer

截至 2026-06-15，Figure AI 是公开资料中最清晰的 humanoid deployment evidence curve 之一：BMW 证明 customer-site measured work，BotQ / Figure 03 证明 manufacturing-process KPI，Catalyst 证明第二客户场景 surface。但 Figure 仍然不是 S5 scaled-commercial-economics proof，因为公开资料还没有披露 robot count by customer、contract value、customer ROI/payback、repeat order、uptime/intervention distribution、robot revenue、gross margin 或 service cost。🟢 Figure official posts dated 2025-11-19 / 2026-04-29 / 2026-05-26; URLs rechecked in prior artifact on 2026-06-15. 🟠 Charlie stage classification.

1. Core question

Q3 不应该问“Figure 是不是赢家”，而应该问：

1. Figure 是否已经把 humanoid 从 demo 推进到可测量的客户现场工作？
2. Figure 02 / Helix / Figure 03 / BotQ 是否构成一条从 deployment proof 到 fleet production 的证据曲线？
3. BMW 的现场 KPI 能否复制到 Catalyst 和其他客户？
4. BotQ 的 production cadence 是否能转换成 customer-utilized fleet，而不是 build-ahead / demo-fleet risk？
5. 哪些新增披露会把 Figure 从 S4 deployment + manufacturing evidence 升级为 S5 economics？

2. Evidence curve: BMW -> Helix -> BotQ -> Catalyst

3. What Figure proves

3.1 Customer-site work is now measurable

BMW is Figure's strongest public evidence because it combines five elements that demo videos usually lack:

1. named customer;
2. active production site;
3. deployment duration;
4. runtime / shift data;
5. task quantity and production context.

The key anchors are 11 months, 10-hour shifts Monday-Friday, 90,000+ parts loaded, 1,250+ runtime hours, and contribution to 30,000+ BMW X3 vehicles. This is S4 because it measures work at a customer site. It is not S5 because the public evidence does not disclose robot count, contract value, customer ROI/payback, repeat order, intervention rate, or margin. 🟢/🟠

3.2 Manufacturing process has become a real evidence unit

BotQ / Figure 03 matters because the disclosed numbers are not just future capacity language. They are production-process KPIs: 350+ Figure 03 delivered, cadence from 1/day to 1/hour, >80% EOL first-pass yield, 99.3% battery-line first-pass yield, 500+ battery packs, 9,000+ actuators, 50+ in-process inspection points, and 80+ functional tests per robot. 🟢

This proves Figure is trying to industrialize humanoid production. It does not prove the produced robots are economically utilized by customers. Production output, delivered robots, contracted deployment, field utilization, recognized revenue, and gross margin remain separate evidence units. 🟢/🟠

3.3 Figure is forming a deployment loop, not just a robot body

Figure's fleet-management language matters because it points to a deployment operating system: real-time health, location, operational status, OTA updates, and field-failure feedback to engineering. 🟢 for company claim.

The commercial implication is still unproven: the public sources do not quantify whether this loop lowers intervention rate, reduces deployment setup time, improves task success, lowers maintenance cost, or improves gross margin. 🟠

4. What Figure does not prove

Figure's public evidence still does not prove S5. The missing data are not cosmetic; each missing item blocks a different part of the economics bridge.

5. Deployment engineering is now the bottleneck

The most useful Q3 framing is: Figure has shifted the question from “can a humanoid perform tasks?” to “can a humanoid company repeatedly deploy fleets with low enough intervention and service burden that customers and vendors both make money?” 🟠

This suggests the bottleneck is no longer only robot body or demo autonomy. The next bottleneck likely sits in deployment engineering:

• site integration;
• task selection and task redesign;
• safety and exception handling;
• maintenance / support workflow;
• intervention reduction;
• customer payback proof;
• fleet learning loop;
• production quality surviving field use.

This is why BMW and BotQ must be read together. BMW tests the customer-site work loop. BotQ tests whether Figure can produce and maintain enough robots to repeat that loop. Catalyst tests whether the loop generalizes beyond one BMW manufacturing use case. 🟠

6. S4-to-S5 upgrade ladder

6.1 Current stage

Current Figure state: S4 deployment + manufacturing evidence.

• BMW: S4 customer-site deployment KPI. 🟢
• BotQ / Figure 03: S4 manufacturing-process KPI. 🟢
• Catalyst: S3/S4 customer expansion surface. 🟢/🟠
• Helix: S3 autonomy/model signal until it is tied to field intervention reduction. 🟢/🟠

6.2 Minimum S5 upgrade bundle

Figure would move materially closer to S5 if at least three of the following are disclosed by primary, customer-side, or filing-quality sources:

1. robot count by customer/site;
2. paid contract value, subscription/revenue model, or recurring economics;
3. customer-confirmed ROI/payback or productivity improvement;
4. repeat order, multi-site rollout, or contract expansion;
5. long-duration uptime and intervention-rate distribution;
6. deployment-linked Figure revenue, gross margin, service cost, or warranty burden;
7. BotQ output tied to contracted customer utilization, not just delivered/internal fleet count.

6.3 Upgrade evidence priority

P0 evidence that would change the thesis fastest:

1. BMW repeat order or second BMW site with robot count and ROI/payback. 🟢 needed.
2. Catalyst follow-through with deployed robot count, task KPI, and rollout timeline. 🟢/🟡 needed.
3. Long-duration intervention-rate distribution in production use. 🟢/🟠 needed.
4. Revenue / gross margin / service burden tied to deployed robots. 🟢/🟠 needed.

P1 evidence that strengthens S4 quality:

1. Additional customer-site task KPI outside BMW. 🟢/🟡 needed.
2. BotQ production cost/yield trend over multiple quarters. 🟢/🟠 needed.
3. Evidence that Helix lowers intervention or deployment setup time in BMW / Catalyst-style sites. 🟢/🟠 needed.
4. Fleet-management metrics: number of active robots, OTA cadence, failure categories, mean time to repair. 🟢/🟠 needed.

7. Figure vs Tesla vs Unitree: different evidence curves

Do not rank Tesla, Figure, and Unitree on a single “winner” score. They answer different bottlenecks.

Source basis: existing Tesla / Figure / Unitree public evidence ledger and Figure / Unitree evidence-debt matrix. 🟠 synthesis from 🟢 company sources.

8. Signal vs noise

Signal

• Named industrial deployment KPI with duration, runtime, task count, and production context. BMW is S4. 🟢
• Manufacturing KPI with cadence, yield, components, battery packs, inspection points, and robot functional tests. BotQ is S4. 🟢
• A second named commercial customer surface with a starting site. Catalyst is S3/S4 until deployment KPI appears. 🟢/🟠
• Fleet-health / OTA / field-failure feedback language. Potential S4 enabler, but economics unproven. 🟢/🟠

Noise unless upgraded

• More autonomy demos without long-duration field success distribution. 🟢 for claim / 🟠 for economics.
• Customer logos without robot count, contract value, task KPI, or repeat order. 🟢/🟠.
• Production cadence interpreted as economically utilized fleet. 🟠.
• “350+ delivered” interpreted as revenue or customer deployment without sell-through/utilization evidence. 🟠.
• Supplier or public-market proxy rumors inferred from Figure production without named customer proof. 🔴/🟠.

9. Common misconceptions

1. Misconception: “Figure 有 BMW KPI，所以 humanoid 商业化已经完成。”

   • Correction: BMW KPI 是强 S4 deployment evidence，但 S5 需要 robot count、customer ROI/payback、repeat order、intervention rate、contract value、margin 和 service cost。🟢/🟠

2. Misconception: “350+ Figure 03 delivered 就等于商业化规模。”

   • Correction: 350+ delivered 是 production-process signal；商业化还需要 sell-through、field utilization、customer economics、revenue 和 gross margin。🟢/🟠

3. Misconception: “Helix no-teleop demo 说明现场 autonomy 已解决。”

   • Correction: 4-minute no-teleop demo 是技术/模型信号；现场 autonomy 需要长时段、多任务、多客户、低 intervention-rate evidence。🟢/🟠

4. Misconception: “Catalyst commercial agreement 就等于收入质量。”

   • Correction: commercial agreement 比普通 logo 强，但没有 robot count、task KPI、deployment duration、contract economics 或 repeat expansion 时，仍不能当 S5。🟢/🟠

5. Misconception: “Figure 私有公司证据可以直接映射到某个公开股票。”

   • Correction: Figure 的证据可能外溢到 compute、components、integrators、customers 或无干净 proxy；不能用 supplier rumor 或 thematic exposure 替代客户/收入验证。🟠

10. Think Deeper questions

1. BMW KPI 如果只能补一个变量，最改变 thesis 的是 robot count、intervention rate、ROI/payback、repeat order，还是 contract value？
2. BotQ 的 1 robot/hour cadence 如果持续，最先应该看 sell-through、field utilization、failure rate，还是 gross margin？
3. Catalyst 是 Figure deployment playbook 的复制，还是另一个定制化试点？什么数据能区分？
4. Figure 的 fleet feedback loop 是 data moat，还是 humanoid deployment 的基本运维成本？
5. 如果 Figure 成为强 S5 私有公司，public-market value capture 更可能在哪一层：compute、components、integrators、customers，还是没有干净表达？
6. Figure 的 bottleneck 到底是 AI/autonomy、hardware reliability、manufacturing yield、deployment service，还是 customer workflow redesign？

11. Public-safe site draft

Figure AI: the clearest deployment curve, still missing S5 economics

Figure 的公开证据价值不在于“它赢了”，而在于它把 humanoid robotics 的两个关键问题变得可测量。

第一是客户现场。Figure 披露 BMW Spartanburg 的 Figure 02 部署：11 months、10-hour shifts Monday-Friday、90,000+ parts loaded、1,250+ runtime hours，并贡献到 30,000+ BMW X3 vehicles。这个证据比 demo video 强，因为它有 named customer、真实生产线、运行时长和任务数量。🟢

第二是制造爬坡。Figure 披露 BotQ delivered 350+ Figure 03 robots，产线 cadence 从 1/day 提到 1/hour，并给出 >80% EOL first-pass yield、99.3% battery-line first-pass yield、9,000+ actuators、50+ in-process inspection points 和 80+ functional tests per robot。这个证据说明 humanoid 公司开始披露生产过程 KPI，而不只是未来愿景。🟢

但结论必须克制：Figure 仍未公开 robot count by customer、contract value、customer ROI/payback、repeat order、uptime/intervention distribution、robot revenue、gross margin 或 service cost。因此 Figure 当前是 S4 deployment + manufacturing evidence，不是 S5 scaled-commercial-economics proof。🟢/🟠

The next question is deployment engineering: can Figure repeat the BMW-style work loop across customers, with low enough intervention and service burden that both the customer and the vendor make money?

Footer: Evidence map only. No winner ranking. No trade recommendation. Deployment KPI and manufacturing KPI are stronger than demos, but they are not full customer economics.

12. Slide-ready compression

Title: Figure AI: strongest deployment curve, still pre-S5 economics

Three-card layout:

1. Customer-site KPI

   • BMW: 11 months; 10-hour shifts Monday-Friday; 90,000+ parts; 1,250+ runtime hours; 30,000+ X3 contribution. 🟢
   • Signal: S4 deployment KPI.

2. Manufacturing KPI

   • BotQ: 350+ Figure 03; 1/day -> 1/hour; >80% EOL FPY; 99.3% battery-line FPY; 9,000+ actuators. 🟢
   • Signal: S4 production-process evidence.

3. Missing before S5

   • Robot count, contract value, ROI/payback, repeat order, intervention rate, revenue, gross margin, service cost. 🟢/🟠
   • Signal: evidence debt, not failure.

Footer: Evidence map only. No trade recommendation. BMW KPI ≠ full economics; production cadence ≠ utilized fleet.

13. Source list

• Figure, “F.02 Contributed to the Production of 30,000 Cars at BMW”, dated 2025-11-19; URL rechecked in robotics-figure-unitree-evidence-debt-matrix-v1.md on 2026-06-15: https://www.figure.ai/news/production-at-bmw 🟢
• Figure, “Introducing Helix 02: Full-Body Autonomy”, dated 2026-01-27; cited from robotics-figure-history-product-deployment-timeline-v1.md: https://www.figure.ai/news/helix-02 🟢
• Figure, “Ramping Figure 03 Production”, dated 2026-04-29; URL rechecked in robotics-figure-unitree-evidence-debt-matrix-v1.md on 2026-06-15: https://www.figure.ai/news/ramping-figure-03-production 🟢
• Figure, “Figure Signs Agreement with Catalyst Brands to Scale Humanoid Operations”, dated 2026-05-26; URL rechecked in robotics-figure-unitree-evidence-debt-matrix-v1.md on 2026-06-15: https://www.figure.ai/news/figure-signs-agreement-with-catalyst-brands 🟢 for announcement / 🟠 economics undisclosed
• Internal cross-reference: knowledge/robotics-figure-history-product-deployment-timeline-v1.md, dated 2026-06-14. 🟠 synthesis from primary Figure sources.
• Internal cross-reference: knowledge/robotics-figure-unitree-s4-to-s5-conversion-dashboard-v1.md, dated 2026-06-14. 🟠 synthesis from primary Figure / Unitree sources.
• Internal cross-reference: knowledge/robotics-figure-unitree-evidence-debt-matrix-v1.md, dated 2026-06-15. 🟠 evidence-debt classification from primary company sources.

14. Public-safety flag

Public-safe: yes, if used as an evidence map.

Do not include:

• Hugo private portfolio data.
• Private channel checks.
• Paid-report excerpts.
• Supplier rumors.
• Buy / sell / hold language.
• Winner ranking.
• Any claim that Figure has proven scaled humanoid economics.
• Any claim that BMW KPI alone proves customer ROI/payback, repeat order, low intervention rate, vendor margin, or service-cost sustainability.