Why robotics now — catalyst evidence cards for morning review

Date: 2026-06-16 Owner: Hugo / Genius Team Agent: Finance / Charlie AGT-002 Status: public-safe research artifact for morning review Visibility: PUBLIC Target use: slide 3 / site section /robotics/why-now/ Related files:

knowledge/robotics-why-now-catalysts-v1.md
knowledge/robotics-why-now-hardware-cost-catalyst-v1.md
knowledge/robotics-tesla-optimus-public-evidence-card-2026-06-16.md
knowledge/robotics-figure-unitree-warrior-brief-v1.md
knowledge/robotics-oem-evidence-curves-beyond-tesla-figure-unitree-v1.md
knowledge/robotics-public-claim-bank-v1.md

Public-safety: yes. No Hugo private portfolio data. No trade recommendation. No paid-report excerpts. No supplier rumor presented as fact.

0. One-line answer

截至 2026-06-16，机器人“why now”的高质量答案不是 TAM 更大，而是证据类型改变：AI/robot foundation model 工具链、低价可获得 humanoid 硬件、真实客户现场 KPI、制造/产能指标、以及 filing-backed 供应链/整机收入开始同时出现。它们合起来说明 humanoid / embodied robotics 正从 demo watching 进入 S3/S4 evidence-tracking phase；但还没有进入 S5 scaled-commercial-economics phase，因为公开资料仍普遍缺 robot count、uptime/intervention distribution、repeat-order economics、customer ROI/payback、gross margin 和 service/warranty burden。🟢 primary-source anchors through 2026-06-16; 🟠 Charlie synthesis.

1. Morning-review framing

Use this as the concise slide/site layer:

Why robotics now? Because the observable variables changed: model stack, body cost, deployment KPI, production KPI, and filing evidence. The investable question is no longer “can robots demo?” but “which evidence chain reaches repeatable economics first?”

Do not frame it as:

“Humanoid robots are solved.” 🔴 unsupported.
“TAM is big, so now is the time.” 🟠 incomplete.
“Tesla / Figure / Unitree / UBTECH / suppliers are buys.” 🔴 outside scope; no trade recommendation.

2. Five catalyst cards

Catalyst	Dated public evidence	What changed	What is still missing	Source grade	Signal
1. AI / robot foundation model stack became a developer platform	NVIDIA Isaac GR00T page describes an open reference platform for general-purpose humanoid robots: open data and data pipelines, open robot foundation model, simulation frameworks, middleware, CUDA-X runtime libraries, and Jetson Thor for real-time robot inference/control. Accessed 2026-06-09 in prior artifact.	Robotics AI moved from isolated demo narrative toward reusable development stack: data, simulation, teleoperation, model, inference and deployment workflow.	Cross-embodiment customer proof, intervention-rate decline, deployment-cost reduction, and paid software economics.	🟢 NVIDIA official page; 🟠 commercialization implication	S3/S4 platform-enablement signal
2. Humanoid body cost / access curve became visible	Unitree official pages list R1 AIR from US$4,900 / R1 from US$5,900, G1 from US$13.5K, H2 at US$29,900; H2 Plus / G1-D pages connect hardware to NVIDIA/Isaac and data/model tooling. Accessed 2026-06-10 in prior artifacts.	A physical humanoid body is no longer only a six-figure lab object; more developers/labs/integrators/customers can experiment.	Humanoid shipments by product line, industrial uptime, customer repeat orders, gross margin, warranty/support burden, software/service revenue.	🟢 Unitree official product pages; 🟠 value-migration implication	S3/S4 cost-access signal
3. Deployment evidence moved beyond viral demo	Figure disclosed BMW Group Plant Spartanburg deployment: 11 months, 10-hour shifts Monday-Friday, 90,000+ parts loaded, 1,250+ runtime hours, contribution to 30,000+ BMW X3 vehicles. Official post dated 2025-11-19. Figure disclosed Catalyst Brands commercial agreement dated 2026-05-26.	Real customer-site KPI appeared: duration, runtime, task output and production-context contribution.	Robot count at customer sites, contract value, customer-confirmed ROI/payback, uptime/intervention distribution, repeat order, gross margin.	🟢 Figure official posts; 🟠 economics classification	S4 deployment signal, not S5
4. Manufacturing / capacity intent became quantifiable	Figure BotQ post dated 2026-04-29: 350+ Figure 03 robots delivered, cadence from 1/day to 1/hour, >80% EOL first-pass yield, 99.3% battery-line FPY, 500+ battery packs, 9,000+ actuators across 10+ SKUs. Tesla Q1 2026 Form 8-K filed 2026-04-22: Fremont first-generation Optimus line designed for 1M robots/year; Texas second-generation line designed for long-term annual capacity of 10M robots.	Manufacturing discussion shifted from “soon mass production” to specific output/cadence/yield/capacity anchors.	Accepted units, actual quarterly output, utilization, line yield over time, customer demand, cost curve, inventory/service quality, revenue/margin.	🟢 Figure official post; 🟢 Tesla SEC filing; 🟠 comparison synthesis	S4 manufacturing / capacity-intent signal
5. Filing-backed commercialization evidence started to appear	UBTECH 2025 annual report disclosed 2025 revenue RMB 2.001bn (+53.3% YoY), full-size embodied humanoid products/services revenue RMB 820.557m (+2,203.7% YoY; 41.1% of revenue), 1,079 full-size embodied humanoid units sold, and >6,000 annualized full-size humanoid capacity at end-2025. 绿的谐波 2025 annual report disclosed revenue RMB 570.714m (+47.31% YoY), net profit RMB 124.367m (+121.42% YoY), harmonic reducer sales 425,158 units (+72.48%), and industrial/embodied-intelligent robot component revenue RMB 422.528m with 34.88% gross margin.	Evidence began entering listed-company filings, which is stronger than concept-stock mapping or anonymous supply-chain checks.	Humanoid-specific gross margin / customer ROI / uptime / intervention / repeat-order proof for UBTECH; named humanoid OEM customer / order / value-content proof for 绿的谐波.	🟢 UBTECH 2025 annual report; 🟢 绿的谐波 2025 annual report; 🟠 sector implication	S4 filing-backed commercial / supply-chain signal

3. Signal / noise compression

Signal

Primary-source KPI replacing demo-only claims: runtime hours, task count, production contribution, delivered robot count, first-pass yield, revenue, unit sales, gross margin. 🟢
Repeated dated anchors across 2025–2026 rather than one-off product launch: Figure BMW 2025-11-19, Figure BotQ 2026-04-29, Tesla Q1 2026 8-K filed 2026-04-22, UBTECH 2025 annual report disclosed 2026-04-15, 绿的谐波 2025 annual report dated/disclosed 2026-04. 🟢
Evidence-chain diversity: AI stack + hardware cost + deployment + manufacturing + filings. 🟠 synthesis.

Noise unless upgraded

Viral videos without robot count, duration, intervention, task KPI, or customer site. 🔴/🟠
Capacity treated as current production. Tesla’s own capacity-table caveat says installed capacity is not current production rate. 🟢/🟠
Customer logos / partnerships without economics. 🟢/🟠
Low price treated as reliability, autonomy or gross-margin proof. 🟢/🟠
Supplier lists / BOM multiplication without confirmed customer, ASP, volume and margin. 🔴/🟠

4. What would turn “why now” into S5 proof

The sector moves from S3/S4 evidence tracking toward S5 scaled-commercial-economics only if multiple companies disclose or customers independently confirm:

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

5. Common misconceptions

Misconception 1: “AI progress means robotics commercialization is automatic.”

Correction: AI stack progress is an enabling catalyst, not commercial proof. Deployment still needs reliability, safety, intervention-rate decline, integration cost, maintenance, payback and repeat orders. 🟠

Misconception 2: “A cheap humanoid proves a winning business model.”

Correction: Unitree-style low prices expand the experiment surface, but value capture may migrate to compute, models, data, deployment services or customer productivity; low price can also compress hardware gross margin. 🟢/🟠

Misconception 3: “Capacity equals production.”

Correction: Figure’s delivered robots / cadence / yield are production-process evidence; Tesla’s 1M/year and 10M/year lines are designed-capacity / long-term-capacity evidence. Both matter, but neither alone proves utilized profitable fleet economics. 🟢/🟠

Misconception 4: “Filing-backed revenue means S5 is here.”

Correction: UBTECH and 绿的谐波 filings are stronger than demos and rumors, but S5 requires revenue quality, customer economics, repeatability, margin durability and service burden. 🟢/🟠

6. Public-site draft section

Why robotics now: the evidence changed, not just the story

The strongest reason to revisit robotics in 2026 is not that the TAM got larger. The stronger reason is that the evidence is becoming measurable.

Five things now appear in public sources at the same time. NVIDIA and others are turning robot AI into a stack of data, simulation, model and inference tools. Unitree is publishing humanoid prices from US$4,900 to US$29,900. Figure is disclosing customer-site runtime and manufacturing KPIs. Tesla is putting Optimus capacity intent into SEC-filed quarterly updates. UBTECH and 绿的谐波 are putting humanoid / embodied-robotics-related revenue, unit and margin evidence into listed-company filings.

That does not mean humanoid robotics has crossed into mature commercial economics. It means the research problem changed. The right question is no longer “can a robot do a viral demo?” The right question 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.

7. Think Deeper questions

If the first S5 proof comes from internal factory use rather than external robot sales, how should ROI be measured: labor substitution, throughput, safety, quality, or uptime-adjusted payback?
If low-cost humanoids commoditize hardware faster than expected, where does value migrate: robot OEM, model/data layer, compute platform, integrator/service layer, or customer productivity capture?
Which filing metric should matter most first: humanoid revenue, robot unit sales, humanoid gross margin, service/warranty cost, or operating cash conversion?
If deployment KPIs improve but gross margin remains weak, is robotics an attractive OEM business or mainly a supplier/software/service opportunity?
What evidence would make this a real return cycle rather than another demo-to-hype cycle?

8. Source list

Primary sources / captured in prior knowledge artifacts:

NVIDIA Isaac GR00T official developer page: https://developer.nvidia.com/isaac/gr00t 🟢 accessed 2026-06-09 in robotics-why-now-catalysts-v1.md.
Figure, “F.02 Contributed to the Production of 30,000 Cars at BMW”, 2025-11-19: https://www.figure.ai/news/production-at-bmw 🟢 captured in robotics-figure-unitree-warrior-brief-v1.md.
Figure, “Ramping Figure 03 Production”, 2026-04-29: https://www.figure.ai/news/ramping-figure-03-production 🟢 captured in robotics-figure-unitree-warrior-brief-v1.md.
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 🟢 captured in robotics-figure-unitree-warrior-brief-v1.md.
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 🟢 captured in robotics-tesla-optimus-public-evidence-card-2026-06-16.md.
Unitree R1 / G1 / H2 / H2 Plus / G1-D official product pages 🟢 accessed 2026-06-10 in robotics-why-now-hardware-cost-catalyst-v1.md and robotics-figure-unitree-warrior-brief-v1.md.
UBTECH Robotics Corp Ltd Annual Report 2025, disclosed 2026-04-15 🟢 captured in robotics-oem-evidence-curves-beyond-tesla-figure-unitree-v1.md.
绿的谐波 2025 annual report PDF: https://static.sse.com.cn/disclosure/listedinfo/announcement/c/new/2026-04-23/688017_20260423_EU45.pdf 🟢 captured in robotics-why-now-catalysts-v1.md and related Leaderdrive artifacts.
Charlie cross-source catalyst synthesis and S3/S4/S5 stage classification, 2026-06-16. 🟠

9. Public-safety exclusions

Do not include:

Hugo private portfolio data, weights, watchlist sizing or trade rationale.
Buy / sell / hold language for TSLA, UBTECH, 绿的谐波 or any robotics-linked company.
Private channel checks, paid-report excerpts, or unverified supplier/customer rumors.
Claims that any one catalyst proves S5 scaled commercial economics.
Claims that designed capacity equals actual output, low price equals reliability, or customer logo equals paid economics.