Robotics morning review decision artifact — Why now evidence ladder v1

Date: 2026-06-13 Owner: Hugo / Genius Team Research owner: Finance / Charlie AGT-002 Status: public-safe research artifact for morning review Visibility: PUBLIC Related files:

• knowledge/robotics-research-harness.md
• knowledge/robotics-sharing-backlog.md
• knowledge/robotics-codex-handoff.md
• knowledge/robotics-why-now-ai-model-catalyst-v1.md
• knowledge/robotics-why-now-hardware-cost-catalyst-v1.md
• knowledge/robotics-why-now-deployment-kpi-catalyst-v1.md
• knowledge/robotics-tesla-optimus-filing-evidence-audit-v1.md
• knowledge/robotics-leaderdrive-case-study-slide-evidence-v3.md

Public-safety: yes. No Hugo private portfolio data. No trade recommendation.

0. One-line answer

截至 2026-06-13，机器人板块的 “why now” 最好不要讲成一个宏大叙事，而应讲成一条证据阶梯：AI/model/data stack 变得可跟踪，低价硬件降低实验门槛，客户现场 KPI 开始出现，制造/产能意图进入 filings，供应链收入/销量开始被报表验证；但整条链仍是 S3/S4-heavy，不是 S5 scaled-commercial-economics proof。

Source grade: 🟢 primary sources from NVIDIA / Physical Intelligence / Skild AI / Unitree / Figure / Agility / UBTECH / Tesla SEC filings / 绿的谐波 filings; 🟠 Charlie synthesis from existing source-backed artifacts as of 2026-06-13.

1. Core question

早晨 review 里要回答的不是 “机器人是不是大机会”，而是：

为什么现在值得更认真地跟踪机器人？哪些证据已经从噪音升级为信号？哪些证据仍然不能支持商业化完成？

当前答案：值得认真跟踪，是因为 evidence unit 变了；但不能升级为 S5，是因为缺 customer-confirmed ROI/payback、低 intervention rate、uptime、repeat order、gross margin、service cost 和 material revenue durability。

2. Evidence ladder: from why-now signal to S5 proof

3. The slide-ready answer

Title: Why robotics now: the evidence unit changed

Subtitle: Not “robots are solved,” but “the industry finally has trackable signals beyond demos.”

Five-card layout:

1. AI stack became trackable

   • NVIDIA GR00T, PI π0/openpi/π0.5/π0.7, Skild industrial-channel/data assets. 🟢
   • Missing: software revenue, deployment economics, attach rate. 🟠

2. Hardware got cheaper and more accessible

   • Unitree R1 AIR from US$4,900; G1 from US$13.5K; H2 US$29,900. 🟢
   • Missing: shipments, reliability, support cost, margin, ROI. 🟠

3. Deployment KPI appeared

   • Figure BMW: 1,250+ runtime hours and 90,000+ parts loaded. Agility GXO: 100,000+ totes moved. 🟢
   • Missing: customer-confirmed payback, intervention rate, contract economics. 🟠

4. Manufacturing / capacity numbers appeared

   • Figure BotQ: 350+ robots delivered, 1/day to 1/hour cadence, >80% EOL FPY. Tesla: Fremont 1M/year designed line, Texas 10M/year long-term designed capacity. 🟢
   • Missing: actual Tesla output, accepted units, yield, robot revenue, gross margin. 🟠

5. Filing-backed revenue / supplier evidence appeared

   • UBTECH: RMB 820.557m full-size embodied humanoid revenue and 1,079 units sold in 2025. 绿的谐波: RMB 422.528m “工业及具身智能机器人零部件” revenue and 425,158 harmonic reducers sold in 2025. 🟢
   • Missing: humanoid-specific margin, customer ROI, named OEM proof, repeat-order economics. 🟠

Required footer: Evidence map only. No winner ranking. No trade recommendation. S4-heavy evidence does not equal S5 scaled-commercial economics.

4. Signal vs noise for morning review

Signal

Noise unless upgraded

5. What would upgrade the whole sector from S4-heavy to S5

The sector-level S5 checklist should require 4 of 6 evidence types across multiple companies or one dominant platform:

1. Paid deployments with repeat orders, robot counts, and contract value. 🟢 required.
2. Customer-confirmed ROI/payback or productivity improvement, not just vendor KPI. 🟢 required.
3. Uptime, intervention rate, safety incident rate, task success rate, and service burden. 🟢 required.
4. Manufacturing output, accepted deliveries, yield, and cost curve. 🟢 required.
5. Humanoid-specific revenue, gross margin, warranty/service cost, and cash-flow contribution. 🟢 required.
6. Supplier-level named customer / order / revenue / margin linkage, not just product fit. 🟢 required.

As of 2026-06-13, reviewed public evidence has meaningful S3/S4 anchors for 1-5, but not enough complete S5 proof.

6. Public site draft section

Why robotics now: not hype, but better evidence

The reason robotics deserves renewed attention is not that humanoids have suddenly become solved products. The reason is that the evidence has improved.

In earlier cycles, investors often had to work from demo videos, customer logos, and TAM slides. By 2026, the evidence base is becoming more concrete: AI/model/tooling stacks are public enough to track; low-cost humanoid hardware gives developers and researchers more access; customer-site KPIs are beginning to appear; manufacturing and capacity intent are being disclosed; and some public-company filings now show humanoid revenue, unit sales, supplier revenue, and component volume.

That is a real shift. But it is not the end of the research. The sector still needs S5 evidence: repeat paid deployment, customer-confirmed ROI, low intervention rate, uptime, service cost, gross margin, and durable revenue. The right framing is therefore: robotics has moved from demo-only noise toward S4-heavy evidence tracking, but the return cycle is still unproven.

Footer: Evidence map only. No winner ranking. No trade recommendation. S4-heavy evidence does not equal S5 scaled-commercial economics.

7. Common misconceptions

1. Misconception: “Why now means commercialization is solved.”

   • Correction: Why-now evidence explains why the research should be upgraded from casual monitoring to disciplined KPI tracking. It does not prove return-cycle economics. 🟠

2. Misconception: “Tesla 1M / 10M capacity means Tesla is already producing those robots.”

   • Correction: Tesla filings use designed-capacity / long-term-capacity language and include the caveat that installed capacity is not current production rate. 🟢

3. Misconception: “Figure / Agility customer KPI equals customer ROI.”

   • Correction: Runtime and task count are strong S4 evidence, but ROI/payback, intervention rate, contract value, and gross margin are still missing. 🟢/🟠

4. Misconception: “Unitree low price proves market adoption.”

   • Correction: Price lowers experiment cost. It does not prove shipments, reliability, support cost, or margin. 🟢/🟠

5. Misconception: “绿的谐波 filing revenue proves specific humanoid OEM relationships.”

   • Correction: Filing-backed revenue and shipment growth are real signals, but named OEM, humanoid-specific revenue, contract value, and margin are not disclosed. 🟢/🟠

8. Think Deeper questions

1. Which metric should become the minimum disclosure standard for humanoids: runtime hours, intervention rate, customer payback, repeat order, or gross margin?
2. If robot bodies get cheaper, does value migrate to OEM scale, model/data platforms, deployment integrators, or suppliers with architecture-specific bottlenecks?
3. Will the first S5 proof come from a private OEM deployment, a public-company filing like UBTECH, an internal Tesla factory KPI, or a supplier order disclosure?
4. If deployment is delivered as RaaS / managed service, how much gross margin is consumed by maintenance, teleoperation, safety, and integration?
5. Which evidence would falsify the current optimism fastest: weak repeat orders, high intervention rate, poor uptime, negative service margin, or customer ROI silence?

9. Source list

Primary / official sources:

• NVIDIA Isaac GR00T official developer page: https://developer.nvidia.com/isaac/gr00t 🟢 accessed 2026-06-09 in existing knowledge base.
• Physical Intelligence, “π0: Our First Generalist Policy,” 2024-10-31: https://www.pi.website/blog/pi0 🟢
• Physical Intelligence, “Open Sourcing π0,” 2025-02-04: https://www.pi.website/blog/openpi 🟢
• Physical Intelligence, “π0.5: a VLA with Open-World Generalization,” 2025-04-22: https://www.pi.website/blog/pi05 🟢
• Physical Intelligence, “π0.7: a Steerable Model with Emergent Capabilities,” 2026-04-16: https://www.pi.website/blog/pi07 🟢
• Skild AI Series A announcement, 2024-07-09: https://www.skild.ai/blogs/announcing-our-300m-series-a 🟢
• Skild AI Series C announcement, 2026-01-14: https://www.skild.ai/blogs/series-c 🟢
• Skild AI ABB / Universal Robots / NVIDIA partnership, 2026-03-19: https://www.skild.ai/blogs/reindustrial-revolution 🟢
• Skild AI Zebra robotics acquisition, 2026-04-15: https://www.skild.ai/blogs/skild-zebra 🟢
• Unitree R1 product page: https://www.unitree.com/R1 🟢 accessed 2026-06-10 in existing knowledge base.
• Unitree G1 product page: https://www.unitree.com/g1/ 🟢 accessed 2026-06-10 in existing knowledge base.
• Unitree H2 product page: https://www.unitree.com/H2 🟢 accessed 2026-06-10 in existing knowledge base.
• Unitree H2 Plus product page: https://www.unitree.com/H2plus 🟢 accessed 2026-06-10 in existing knowledge base.
• Unitree G1-D product page: https://www.unitree.com/G1-D 🟢 accessed 2026-06-10 in existing knowledge base.
• Figure, “F.02 Contributed to the Production of 30,000 Cars at BMW,” 2025-11-19: https://www.figure.ai/news/production-at-bmw 🟢 checked 2026-06-10.
• Figure, “Ramping Figure 03 Production,” 2026-04-29: https://www.figure.ai/news/ramping-figure-03-production 🟢 checked 2026-06-10.
• Agility Robotics, GXO multi-year agreement, 2024-06-27: https://agilityrobotics.com/content/gxo-signs-industry-first-multi-year-agreement-with-agility-robotics 🟢 checked 2026-06-10.
• Agility Robotics, Digit 100,000+ totes milestone, 2025-11-20: https://agilityrobotics.com/content/digit-moves-over-100k-totes 🟢 checked 2026-06-10.
• UBTECH Robotics Corp Ltd Annual Report 2025, disclosed 2026-04-15: https://owebsite-cdn.ubtrobot.com/resources/file/2026/04/15/795097683329093.pdf 🟢 checked 2026-06-11.
• Tesla Q4 2025 Update, Form 8-K Exhibit 99.1, filed 2026-01-28: https://www.sec.gov/Archives/edgar/data/1318605/000162828026003837/exhibit991.htm 🟢 checked 2026-06-12.
• Tesla Q1 2026 Update, Form 8-K Exhibit 99.1, filed 2026-04-22: https://www.sec.gov/Archives/edgar/data/1318605/000162828026026551/exhibit991.htm 🟢 checked 2026-06-12.
• 绿的谐波 2025 annual report, disclosed 2026-04-23: https://static.sse.com.cn/disclosure/listedinfo/announcement/c/new/2026-04-23/688017_20260423_EU45.pdf 🟢 checked 2026-06-11.
• 绿的谐波 2026-06-09 abnormal trading announcement: https://static.sse.com.cn/disclosure/listedinfo/announcement/c/new/2026-06-09/688017_20260609_MZU2.pdf 🟢 checked 2026-06-11.

Internal synthesis sources:

• knowledge/robotics-why-now-ai-model-catalyst-v1.md 🟠
• knowledge/robotics-why-now-hardware-cost-catalyst-v1.md 🟠
• knowledge/robotics-why-now-deployment-kpi-catalyst-v1.md 🟠
• knowledge/robotics-tesla-optimus-filing-evidence-audit-v1.md 🟠
• knowledge/robotics-leaderdrive-case-study-slide-evidence-v3.md 🟠

10. Public-safety flag

Public-safe if used as an evidence map and not as a recommendation.

Do not publish:

• Hugo private portfolio data.
• Buy / sell / hold language.
• Winner ranking.
• Claims that S4 signals equal S5 commercialization.
• Claims that designed capacity equals current production.
• Supplier-customer inferences without named customer, order, filing, teardown, or revenue evidence.