Investment ResearchTopic research, evidence curves, and cycle signals
HomeSign inEN

Research library · updated 2026-06-14 · public

Figure AI / Unitree S4-to-S5 conversion dashboard v1

Date: 2026-06-14 Owner: Hugo / Genius Team Agent: Finance / Charlie AGT-002 Status: public-safe research artifact Visibility: PUBLIC Primary use: morning review, slide 11, /robotics/companies, /robotics/signals

Related files:

  • knowledge/robotics-research-harness.md
  • knowledge/robotics-figure-unitree-evidence-v1.md
  • knowledge/robotics-figure-unitree-public-compression-guardrail-v1.md
  • knowledge/robotics-tesla-figure-unitree-public-evidence-ledger-v1.md
  • knowledge/robotics-figure-history-product-deployment-timeline-v1.md

Public-safety note:

  • Public-safe: yes.
  • No Hugo private portfolio data.
  • No trade recommendation.
  • No winner ranking.
  • Company official claims are primary evidence for what the company disclosed, not proof of full customer economics.

0. One-line answer

截至 2026-06-14，Figure AI 和 Unitree / 宇树最适合被读成两条不同的 S4-to-S5 转换路径：

  • Figure 的 S4 证据已经较强：BMW 产线部署披露 11 months、10-hour shifts Monday-Friday、90,000+ parts loaded、1,250+ runtime hours、30,000+ BMW X3 vehicles contribution；BotQ 披露 350+ Figure 03 delivered、1/day to 1/hour cadence、EOL first-pass yield >80%、battery-line first-pass yield 99.3%、9,000+ actuators。🟢 Figure official posts dated 2025-11-19 and 2026-04-29; captured in prior evidence pack 2026-06-10.
  • Unitree 的 S3/S4 证据在 cost/access 与 developer-platform：R1 AIR from US$4,900 / R1 from US$5,900、G1 from US$13.5K、H2 US$29,900、H2 Plus with NVIDIA Jetson / Isaac GR00T / TeleOp / Sim references、G1-D data acquisition / model workflow. 🟢 Unitree official product pages captured 2026-06-10.
  • 两者都还缺 S5 证据：repeat paid deployment、robot count by customer、uptime/intervention distribution、customer ROI/payback、support/warranty burden、gross margin / revenue materiality。🟠 Charlie evidence-ladder synthesis as of 2026-06-14.

1. Core question

不是“Figure 和 Unitree 谁赢”，而是：

  1. Figure 的部署 KPI 能否转换成 repeatable paid deployment economics？
  2. Unitree 的低价格和工具链能否转换成 shipment scale、developer adoption、industrial reliability 与可持续 margin？
  3. 哪些新增披露会把 S4 measurable evidence 升级成 S5 scaled-commercial-economics proof？

2. Conversion dashboard

PathCurrent strongest evidenceAs-of dateSource gradeCurrent signalS5 conversion evidence neededWhy this matters
Figure customer-site deploymentBMW Spartanburg: 11-month deployment; 10-hour shifts Monday-Friday; 90,000+ parts loaded; 1,250+ runtime hours; contributed to 30,000+ BMW X3 vehicles.Figure post dated 2025-11-19; captured 2026-06-10🟢 primary company sourceS4 deployment KPICustomer-confirmed ROI/payback; robot count; paid contract value; uptime/intervention distribution; repeat order or multi-site rollout.Converts task KPI from “robot can work” into “customer repeatedly pays because economics work.”
Figure manufacturing processBotQ: 350+ Figure 03 delivered; cadence from 1/day to 1/hour; EOL FPY >80%; battery-line FPY 99.3%; 500+ battery packs; 9,000+ actuators across 10+ SKUs.Figure post dated 2026-04-29; captured 2026-06-10🟢 primary company sourceS4 manufacturing KPISell-through to customers; field utilization of produced units; production cost/yield trend; gross margin; warranty/service burden.Separates production capability from economically utilized fleet.
Figure commercial expansionCatalyst commercial agreement to deploy humanoids in distribution/logistics network, starting at Reno, Nevada Distribution Logistics Center.Figure post dated 2026-05-26; captured 2026-06-10🟢 primary for agreement; 🟠 economics undisclosedS3/S4 customer expansionRobot count; deployment schedule; contract value; workflow KPI; repeat-order terms; site expansion.Tests whether BMW-style evidence generalizes beyond one industrial task/customer.
Unitree hardware access curveR1 AIR from US$4,900 / R1 from US$5,900; G1 from US$13.5K; H2 US$29,900.Unitree pages captured 2026-06-10🟢 primary product pagesS3/S4 cost-accessHumanoid shipments by model; customer mix; reliability/warranty data; gross margin; repeat customer purchases.Low price can expand the experiment base, but S5 requires scale and sustainable economics.
Unitree developer / model workflowH2 Plus references NVIDIA Jetson T5000, FP4 2,070 TFLOPS, Isaac GR00T / TeleOp / Sim; G1-D describes data acquisition, labeling, training and inference workflow.Unitree pages captured 2026-06-10🟢 primary product pagesS3 platform-enablementActive developer count; dataset scale; paid software/tooling revenue; model-performance adoption; deployments using the workflow.Tests whether Unitree captures platform/data value or mainly sells robot bodies.

3. Signal ladder: what upgrades the thesis

3.1 Figure upgrade ladder

S3: official demo or customer partnership without economics.

  • Example: Helix 02 autonomy demos are useful, but by themselves remain demo/autonomy evidence. 🟢 Figure official post dated 2026-01-27; 🟠 commercial implication.

S4: named customer deployment KPI or manufacturing KPI.

  • Current Figure state: BMW runtime/task metrics + BotQ production/yield/cadence metrics. 🟢 as of 2026-06-14 based on prior captures.

S5: repeatable economics and financial materiality.

  • Minimum upgrade bundle: at least 3 of 6 are disclosed by customer/company/filing-quality sources:
    1. robot count by site;
    2. paid contract value or recurring economics;
    3. uptime and intervention rate over long-duration production use;
    4. customer-confirmed productivity gain / ROI / payback;
    5. repeat order or multi-site rollout;
    6. Figure revenue, gross margin, service/support cost tied to deployments.

3.2 Unitree upgrade ladder

S3: official product pricing/specs/tooling.

  • Current Unitree state: public price anchors and H2 Plus / G1-D workflow pages. 🟢 as of 2026-06-10 captures.

S4: shipment scale or named deployment KPI.

  • Upgrade evidence would be official/customer disclosure of humanoid unit shipments by model, named industrial deployments, runtime, or intervention data. 🟢/🟡 needed.

S5: durable commercial economics.

  • Minimum upgrade bundle: at least 3 of 6 are disclosed by official/customer/filing-quality sources:
    1. verified humanoid shipments by model and customer type;
    2. repeat orders / fleet expansion;
    3. industrial uptime and intervention-rate data;
    4. warranty, maintenance, and support burden;
    5. product-line gross margin or service/software attachment;
    6. customer ROI/payback or paid deployment economics.

4. Signal vs noise

Signal

  • Figure BMW runtime/task metrics: S4 because they include named industrial site, duration, shifts, task quantity and runtime. 🟢
  • Figure BotQ manufacturing metrics: S4 because they include delivered robots, cadence, yields, battery packs and actuator output. 🟢
  • Figure Catalyst agreement: S3/S4 because it names a commercial customer and starting site, but lacks economics. 🟢/🟠
  • Unitree official price anchors below US$30K: S3/S4 because they change hardware access cost; S4 only if tied to shipment/adoption evidence later. 🟢/🟠
  • Unitree G1-D / H2 Plus tooling: S3 because it shows data/model workflow packaging, not yet adoption or economics. 🟢/🟠

Noise unless upgraded

  • Viral videos without runtime, task repetition, intervention rate, safety record, or customer context. 🔴/🟠
  • Customer logos without robot count, contract value, task KPI, or repeat order. 🟢/🟠
  • Low price interpreted as proof of industrial reliability or margin. 🟠
  • Production cadence interpreted as customer utilization. 🟢/🟠
  • NVIDIA / Isaac / open-source framework references interpreted as Unitree-owned software economics. 🟢/🟠

5. Public-safe slide/module draft

Figure / Unitree: two S4-to-S5 paths

Figure 的问题是：部署 KPI 能不能变成可重复付费经济性？

截至 2026-06-14，Figure 的强证据是 BMW 和 BotQ。BMW 产线部署披露 11 months、10-hour shifts Monday-Friday、90,000+ parts loaded、1,250+ runtime hours、30,000+ BMW X3 vehicles contribution；BotQ 披露 350+ Figure 03 delivered、1/day to 1/hour cadence、EOL FPY >80%、battery-line FPY 99.3%。这些是 S4，不是 S5，因为还没有 contract value、robot count、customer ROI/payback、intervention rate、repeat order 和 margin。🟢/🟠

Unitree 的问题是：低价格和开发者工具链能不能变成 shipment scale 与可持续价值捕获？

截至 2026-06-14，Unitree 的强证据是 cost/access curve：R1 AIR from US$4,900 / R1 from US$5,900、G1 from US$13.5K、H2 US$29,900；H2 Plus / G1-D 把 robot body 接到 Jetson、Isaac、TeleOp、Sim、data acquisition、model training / inference workflow。但这还不是 S5，因为缺 shipment scale、industrial deployment KPI、support cost、gross margin 和 customer ROI。🟢/🟠

Safe conclusion: Figure measures work; Unitree lowers access cost. Neither yet proves scaled commercial economics.

Footer: Evidence map only. No winner ranking. No trade recommendation. Deployment KPI ≠ full economics; low price ≠ reliability or margin.

6. Common misconceptions

  1. Misconception: “Figure 已经有 BMW 指标，所以 humanoid 商业化完成。” Correction: BMW 指标是强 S4 deployment KPI，但 S5 还需要客户经济性、repeat orders、intervention rate、service cost 与 margin。🟢/🟠

  2. Misconception: “Unitree 价格低，所以一定捕获最大价值。” Correction: 低价可以扩大实验基数，也可能压缩硬件 margin；价值可能迁移到模型、数据、集成、服务或客户流程控制。🟠

  3. Misconception: “生产出来的机器人数量就是商业部署数量。” Correction: production output、delivered robots、field utilization、paid deployment、recognized revenue 是五个不同指标，不能混用。🟢/🟠

  4. Misconception: “开发者工具链等于软件收入。” Correction: 工具链说明平台表面正在形成；S5 仍需付费 adoption、attach rate、revenue、margin 或客户 ROI。🟢/🟠

7. Think Deeper questions

  1. Figure 最小 S5 披露组合应该是什么：客户 ROI、repeat order、intervention rate、还是 gross margin？如果只能等一个，哪一个最改变 thesis？
  2. Unitree 的低价格更可能扩大市场，还是更可能压缩硬件价值捕获？什么数据可以区分这两种路径？
  3. 如果 Figure 证明任务 KPI 但保持私有，public-market 如何避免把 supplier rumor 误读成已验证 exposure？
  4. 如果 Unitree 成为广泛使用的 humanoid developer platform，价值会留在 robot OEM、迁移到 NVIDIA/模型层、还是被 integrator/customer 吸收？
  5. 对整个 sector 来说，下一次 S5 升级更可能来自 customer economics、production yield、shipment scale、还是 safety/compliance gate？

8. What to track next

Figure watch items:

  • BMW or another customer confirms robot count, ROI/payback, or repeat order. 🟢 needed.
  • Catalyst publishes deployment count, task KPI, or rollout schedule. 🟢/🟡 needed.
  • Figure discloses revenue, gross margin, service/support cost, or fleet utilization. 🟢/🟠 needed.
  • Additional BotQ metrics connect production output to customer deployment rather than internal fleet inventory. 🟢/🟠 needed.

Unitree watch items:

  • Humanoid shipments by R1/G1/H2 model and customer type. 🟢 needed.
  • Named industrial/customer deployment KPI with runtime and intervention data. 🟢/🟡 needed.
  • Warranty/maintenance burden and gross margin by humanoid product line. 🟢/🟠 needed.
  • Evidence that H2 Plus / G1-D workflows generate paid software/tooling/service revenue or measurable developer adoption. 🟢/🟡/🟠 needed.

9. Source list

Figure AI:

Unitree / 宇树:

Internal synthesis / control files:

  • knowledge/robotics-figure-unitree-evidence-v1.md, dated 2026-06-10. 🟠 synthesis from primary company pages.
  • knowledge/robotics-figure-unitree-public-compression-guardrail-v1.md, dated 2026-06-13. 🟠 packaging guardrail.
  • knowledge/robotics-tesla-figure-unitree-public-evidence-ledger-v1.md, dated 2026-06-13. 🟠 cross-company ledger.

10. Public-safe flag

Public-safe: yes.

Do not include:

  • Hugo private portfolio data.
  • Private channel information.
  • Paid-report excerpts.
  • Buy / sell / hold framing.
  • Winner ranking.
  • Any claim that Figure has proven full scaled economics.
  • Any claim that Unitree low price proves industrial reliability, shipment scale, gross margin, or customer ROI.