Robotics morning review — evidence-first executive brief

Date: 2026-06-17 Owner: Hugo / Genius Team Agent: Finance / Charlie AGT-002 Status: SYNTHESIS_CANDIDATE Visibility: PUBLIC Target use: morning review / slide-3 opener / /robotics/why-now/ pre-read Public-safety: no Hugo private portfolio data; no trade recommendation; no paid-report excerpts; no private channel checks.

Related source files:

• knowledge/robotics-research-harness.md
• knowledge/robotics-morning-review-why-now-evidence-spine-2026-06-16.md
• knowledge/robotics-why-now-catalyst-evidence-cards-2026-06-16.md
• knowledge/robotics-morning-review-tesla-figure-unitree-evidence-split-2026-06-16.md
• knowledge/robotics-leaderdrive-public-safe-slide-10-case-study-2026-06-16.md
• knowledge/robotics-codex-handoff.md

0. One-line answer

截至 2026-06-17，morning review 最应该保留的一句话是：机器人研究已经从“看 demo”变成“追证据链”，但还没有进入 S5 scaled-commercial-economics phase。最硬的变化是 5 类公开证据同时出现：AI/data stack、低价 humanoid hardware access、真实客户现场 KPI、制造/产能 KPI、filing / fleet denominator。🟢 linked primary-source anchors through 2026-06-16; 🟠 Charlie synthesis.

核心句：

The evidence changed before the economics were proven.

1. Morning-review opening frame

不要把开场写成“机器人一定爆发”或“哪个公司会赢”。更好的投资研究开场是：

1. 过去机器人周期的问题：TAM 大、demo 强，但缺少可复盘的商业化证据。🟠 framework judgment.
2. 2025-2026 的变化：公开资料开始出现 runtime、task count、cadence、yield、price anchor、revenue/unit/gross-margin line items。🟢/🟠 source-backed synthesis.
3. 当前阶段：S3/S4 evidence-tracking phase，不是 S5 scaled-commercial-economics phase。🟠 classification.
4. 下一步问题：哪条证据链先补齐 repeat order、intervention rate、customer ROI/payback、gross margin、service/warranty burden。🟠 falsification framework.

2. Five evidence chains to show first

3. Tesla / Figure / Unitree: evidence split, not horse race

The morning review should not rank Tesla / Figure / Unitree as winners. Use them as three different proof types:

4. Leaderdrive / 绿的谐波: use as supply-chain evidence, not portfolio story

Leaderdrive / 绿的谐波 is useful for the morning review only if sanitized as a public-safe case study:

• Proven: 2025 revenue RMB 570.714m (+47.31% YoY), net profit RMB 124.367m (+121.42%), harmonic reducer sales 425,158 units (+72.48%), industrial/embodied-intelligent robot component revenue RMB 422.528m with 34.88% gross margin. 🟢 绿的谐波 2025 annual report, disclosed 2026-04-23.
• Interpretation: this is stronger than “robot concept stock” narrative because it appears in filings and includes revenue / volume / margin anchors. 🟢/🟠.
• Boundary: reviewed public sources still do not disclose named humanoid OEM customer, humanoid-only revenue/margin, contract value, order volume, single-robot value content, or customer ROI/payback. 🟢/🟠.
• Use-case: “filing-backed supplier signal, not S5 humanoid winner proof.” 🟠.

Do not include Hugo position size, purchase context, private rationale, or any trade language.

5. Signal vs noise

Signal

1. Customer-site runtime and task output: Figure BMW 1,250+ runtime hours and 90,000+ parts. 🟢
2. Manufacturing cadence and yield: Figure BotQ 350+ robots, 1/hour cadence, >80% EOL FPY, 99.3% battery-line FPY. 🟢
3. Filing-backed capacity intent: Tesla Optimus 1M/year Fremont designed line and 10M/year long-term Texas designed capacity. 🟢
4. Public hardware price anchors: Unitree US$4,900-29,900 humanoid price range in reviewed official pages. 🟢
5. Filing-backed revenue/unit/margin line items: UBTECH humanoid revenue/units and 绿的谐波 reducer/component revenue/volume/margin. 🟢
6. Evidence-chain diversity across AI stack, body cost, deployment, manufacturing and filings. 🟠 synthesis.

Noise unless upgraded

1. Viral demo without runtime, task denominator, reset/intervention rate, customer site or safety/maintenance burden. 🔴/🟠
2. Capacity target treated as actual output. 🟠
3. Customer logo or partnership treated as ROI/payback. 🟠
4. Low price treated as reliability, autonomy or margin proof. 🟠
5. Supplier/BOM multiplication without named customer, ASP, volume and margin proof. 🔴/🟠
6. Funding rounds treated as commercialization. 🟡 for funding fact; 🟠 for economics inference.

6. S4-to-S5 upgrade checklist

Upgrade the whole “why robotics now” thesis toward S5 only if at least three of the following become public from primary/company/customer/filing sources:

1. Accepted robots by customer/site and by quarter. 🟢
2. Runtime / uptime / intervention-rate distribution under production conditions. 🟢
3. Repeat orders or multi-site expansion with comparable workflow economics. 🟢
4. Customer ROI/payback, or enough transparent data to estimate payback. 🟢/🟠
5. Robot revenue, gross margin, service/support cost and warranty burden. 🟢
6. Supplier order/revenue exposure with named OEM or filing-backed humanoid-specific split. 🟢

Downgrade if KPI disclosures remain one-off showcases for 12-24 months, production/capacity expands without field utilization disclosure, or low-cost hardware growth brings high service burden and weak customer adoption. 🟠

7. Public-site draft paragraph

The strongest reason to revisit robotics in 2026 is not that the TAM slide became larger. It is that the evidence unit changed. The sector now has public anchors for robot-AI data pipelines, humanoid body pricing, customer-site runtime, manufacturing cadence, designed capacity, service-robot denominators, and filing-backed revenue/unit data. That does not prove scaled humanoid economics. But it changes the research problem: the question is no longer whether robots can produce viral demos; it is which evidence chain reaches customer ROI, low intervention, repeat orders, production yield, revenue quality and durable margin first.

Footer: Evidence map only. No winner ranking. No trade recommendation. S3/S4 evidence is not S5 economics.

8. Common misconception corrections

1. “Why now means humanoids are ready.”

   • Correction: why-now means the evidence became measurable; readiness still requires S5 economics. 🟠

2. “TAM is the main reason.”

   • Correction: TAM is not timing. Timing comes from measurable changes in cost, deployment, manufacturing, data and filings. 🟠

3. “Tesla capacity target equals Tesla production.”

   • Correction: Tesla filing evidence is capacity-intent / manufacturing-infrastructure evidence, not actual output or commercialization proof. 🟢/🟠

4. “Figure BMW proves the sector is solved.”

   • Correction: BMW is high-quality S4 deployment evidence; S5 still needs robot count, uptime/intervention, ROI/payback, repeat order and margin. 🟢/🟠

5. “Unitree low price means winner.”

   • Correction: lower price expands experimentation but may also commoditize hardware and shift value to models, compute, deployment services or customer workflow ownership. 🟢/🟠

6. “Filing-backed supplier revenue means all suppliers benefit.”

   • Correction: filing evidence is stronger than rumor, but supplier value capture still needs named customer, volume, ASP, margin durability and architecture fit. 🟢/🟠

9. Think Deeper questions

1. Which KPI is the first real S5 bridge: customer ROI, intervention-rate decline, repeat orders, humanoid gross margin, or production yield?
2. If low-cost humanoid bodies commoditize hardware, where does value migrate: OEM, AI/model provider, compute platform, integrator, deployment-service layer, component supplier, or customer?
3. Does Tesla’s internal-factory route require a different evidence standard than Figure’s external-customer route?
4. Should Amazon-style mature robot fleet evidence be used as a denominator benchmark for humanoids, or only as a warning against overreading small humanoid deployments?
5. If supplier filings improve but named humanoid OEM exposure remains undisclosed, should the signal stay S4 or be discounted toward S3?

10. Source list

Primary / official / source-backed references captured in linked artifacts:

• NVIDIA Isaac GR00T official developer page and related NVIDIA technical materials captured in why-now artifacts through 2026-06-16. 🟢
• Unitree R1 / G1 / H2 / H2 Plus / G1-D official product pages, accessed 2026-06-10 in prior artifacts. 🟢
• Figure, “F.02 Contributed to the Production of 30,000 Cars at BMW”, 2025-11-19. 🟢
• Figure, “Ramping Figure 03 Production”, 2026-04-29. 🟢
• Figure, “Figure Signs Agreement with Catalyst Brands to Scale Humanoid Operations”, 2026-05-26. 🟢
• Tesla Q1 2026 Update, Form 8-K Exhibit 99.1, filed 2026-04-22. 🟢
• IFR World Robotics 2025 Service Robots press release / executive-summary materials, 2025-10-07, captured in prior artifacts. 🟢
• UBTECH Robotics Corp Ltd Annual Report 2025, disclosed 2026-04-15, captured in prior artifacts. 🟢
• 绿的谐波 2025 annual report PDF, disclosed 2026-04-23, captured in Leaderdrive / why-now artifacts. 🟢
• Charlie cross-source synthesis and S3/S4/S5 classification, 2026-06-17. 🟠

11. Public-safety exclusions

Do not include:

• Hugo private portfolio data, weights, purchase prices, watchlist sizing, or trade rationale.
• Buy / sell / hold language for Tesla, NVIDIA, UBTECH, 绿的谐波, Figure, Unitree or any robotics-linked company.
• Private channel checks, paid-report excerpts, or unverified supplier/customer rumors.
• Claims that any catalyst proves S5 scaled commercial economics by itself.
• Claims that designed capacity equals actual output, low price equals reliability, funding equals commercialization, customer logo equals economics, or filing-backed supplier revenue equals named humanoid design win.