Robotics Morning Review Claim-Check Matrix — Tesla / Figure / Unitree

Date: 2026-06-17 As-of: 2026-06-17 03:49 UTC live URL check for Figure / Unitree; Tesla SEC source uses 2026-06-16 local SEC extraction appendix because live SEC request returned HTTP 403 on 2026-06-17. Owner: Hugo / Genius Team Research owner: Finance / Charlie AGT-002 Status: SYNTHESIS_CANDIDATE Visibility: PUBLIC Output target: morning-review appendix / Codex packaging guardrail for slide 11 or /robotics/companies

Related files:

knowledge/robotics-research-harness.md
knowledge/robotics-morning-review-tesla-figure-unitree-evidence-split-2026-06-16.md
knowledge/robotics-tesla-figure-unitree-public-evidence-ledger-v1.md
knowledge/robotics-tesla-optimus-sec-filing-excerpt-appendix-2026-06-16.md

Public-safety flag: PUBLIC if used as written. Excludes Hugo private portfolio information, private channel checks, supplier rumors, and buy / sell / hold language.

0. One-line answer

截至 2026-06-17，Tesla / Figure / Unitree 的公开证据应被压缩成一个 claim-check matrix，而不是 winner ranking：Tesla 证明的是 filing-backed capacity intent，Figure 证明的是 bounded deployment + manufacturing KPI，Unitree 证明的是 low-cost access + developer tooling；三者都没有公开证明 S5 scaled commercial economics。🟢/🟠

Canonical footer:

Evidence map only. No winner ranking. No trade recommendation. Capacity ≠ production; deployment KPI ≠ economics; low price ≠ reliability/margin.

1. Why this artifact exists

现有材料已经有完整三家公司证据分拆。本文件的增量是给早晨 review / Codex 一个“压缩前的事实核验表”：每个 claim 能写到什么程度、证据等级是什么、as-of 日期是什么、最容易误写成什么。

2026-06-17 live check summary:

Figure BMW / BotQ / Catalyst official pages returned HTTP 200 and matched key BMW / production / Catalyst anchors. 🟢 live check 2026-06-17.
Unitree R1 / G1 / H2 / H2 Plus / G1-D official pages returned HTTP 200 and matched price / tooling anchors. 🟢 live check 2026-06-17.
Tesla Q1 2026 SEC Exhibit 99.1 URL returned HTTP 403 from this environment on 2026-06-17; use the 2026-06-16 local SEC extraction appendix as the current verified Tesla excerpt layer. 🟠 method note; original SEC filing remains 🟢 primary.

2. Claim-check matrix

Company	Public claim Codex can write	Strongest evidence anchor	As-of / source grade	Signal	Do not write
Tesla Optimus	Tesla has SEC filing-backed manufacturing-infrastructure / capacity-intent evidence for Optimus.	Q1 2026 8-K Exhibit 99.1 excerpt in local appendix: Fremont first-generation line designed for 1M robots/year; Gigafactory Texas second-generation line designed for long-term annual capacity of 10M robots.	🟢 Tesla primary filing filed 2026-04-22; 🟠 local extraction appendix dated 2026-06-16; live SEC request returned 403 on 2026-06-17.	S4 infrastructure / capacity-intent	Tesla is producing 1M/year; Tesla has 10M/year capacity today; Tesla disclosed accepted units, yield, runtime, ROI, revenue, gross margin, or named suppliers.
Figure AI — BMW	Figure has unusually concrete public customer-site task KPI in a bounded industrial setting.	Figure BMW post matched 90,000+ parts, 1,250+ runtime hours, 30,000+ BMW X3 contribution, 10-hour shifts, and BMW anchor in live check.	🟢 Figure official post dated 2025-11-19; URL HTTP 200 and anchors matched 2026-06-17.	S4 deployment KPI	BMW KPI proves customer ROI/payback, repeat order, robot count by customer, intervention rate, Figure gross margin, or scaled economics.
Figure AI — BotQ	Figure has unusually concrete manufacturing-process KPI for Figure 03 production ramp.	Figure BotQ post matched 350+ robots, >80% EOL first-pass yield, 99.3% battery-line first-pass yield, and 9,000+ actuators in live check. The phrase “1 robot / hour” should be rechecked manually before exact quotation because the live keyword `1 hour` did not match.	🟢 Figure official post dated 2026-04-29; URL HTTP 200 and core anchors matched 2026-06-17; cadence wording remains 🟠 until exact text is re-opened.	S4 manufacturing KPI	BotQ KPI proves utilized fleet, customer deployment scale, customer acceptance, gross margin, service burden, or repeat commercial economics.
Figure AI — Catalyst	Figure added another named commercial logistics deployment surface.	Figure Catalyst post matched Catalyst, Reno, logistics, and agreement anchors in live check.	🟢 Figure official post dated 2026-05-26; URL HTTP 200 and anchors matched 2026-06-17; economics undisclosed 🟠.	S3/S4 customer-surface signal	Catalyst proves contract value, robot count, paid terms, repeat order, ROI/payback, or revenue materiality.
Unitree R1	Unitree has explicit low-cost humanoid access anchors around the R1 family.	Unitree R1 page matched US$4,900, US$5,900, and R1 anchors in live check.	🟢 Unitree official page; URL HTTP 200 and anchors matched 2026-06-17.	S3/S4 cost-access	Low price proves industrial reliability, autonomy, shipment scale, gross margin, support cost, or customer ROI.
Unitree G1 / H2	Unitree has public humanoid price anchors across lower-cost and higher-capability products.	Unitree G1 page matched US$13.5K and G1; H2 page matched US$29,900 and H2 in live check.	🟢 Unitree official pages; URLs HTTP 200 and anchors matched 2026-06-17.	S3/S4 cost-access	Price list proves deployment economics, repeat orders, warranty burden, or durable hardware margin.
Unitree H2 Plus / G1-D	Unitree product pages support a developer-platform / data-workflow claim, not a full AI-value-capture claim.	H2 Plus matched NVIDIA, Isaac, TeleOp, Sim; G1-D matched Data, model, G1-D in live check.	🟢 Unitree official pages; URLs HTTP 200 and anchors matched 2026-06-17.	S3/S4 developer-platform signal	NVIDIA / Isaac / data workflow references prove Unitree owns the model layer, software monetization, autonomy, or customer economics.

3. Signal vs noise

Signal

Tesla: dated SEC-filed manufacturing-site and designed-capacity language is stronger than demo video evidence. 🟢/🟠 as of 2026-04-22 filing and 2026-06-16 extraction.
Figure BMW: named customer + duration / shifts / runtime / task count / production-context metrics are stronger than a customer logo. 🟢 as of 2025-11-19 post, live checked 2026-06-17.
Figure BotQ: robot count / yield / battery / actuator manufacturing metrics are stronger than “factory planned” language. 🟢 as of 2026-04-29 post, live checked 2026-06-17.
Unitree: official price anchors and product/tooling pages are valid evidence for cost-access and developer experimentation. 🟢 live checked 2026-06-17.
Cross-company classification is useful only as synthesis, not primary fact. 🟠 Charlie synthesis 2026-06-17.

Noise unless upgraded

Designed capacity interpreted as current production or accepted units. 🟠 / 🔴 if stated as fact.
Customer-site KPI interpreted as full unit economics without ROI, robot count, intervention, repeat order, revenue, margin, and service-cost evidence. 🟠 / 🔴 if stated as fact.
Low price interpreted as reliability, high gross margin, industrial readiness, or software value capture. 🟠 / 🔴 if stated as fact.
Supplier rumor attached to Tesla / Figure / Unitree without named OEM confirmation, contract/order value, teardown, or filing-backed revenue exposure. 🔴
Viral demo clips without runtime, reset, intervention, safety, task repeatability, and customer-site context. 🔴

4. Slide / site compression draft

Title: Tesla / Figure / Unitree: three evidence curves, three proof debts

Subtitle: Robotics evidence is now measurable, but still not S5 scaled-commercial economics.

Three-card structure:

Tesla — capacity intent
- Proven: SEC filing-backed Fremont 1M/year designed line and Texas 10M/year long-term designed line. 🟢/🟠 as of 2026-04-22 filing and 2026-06-16 extraction.
- Proof debt: actual output, accepted units, yield, runtime, intervention, factory ROI, external revenue, gross margin. 🔴 not disclosed in reviewed filings.
Figure — deployment + manufacturing KPI
- Proven: BMW 90,000+ parts, 1,250+ runtime hours, 30,000+ X3 contribution; BotQ 350+ robots, >80% EOL FPY, 99.3% battery-line FPY, 9,000+ actuators. 🟢 Figure official posts; live checked 2026-06-17.
- Proof debt: customer-confirmed ROI/payback, repeat orders, robot count by customer, intervention profile, revenue, gross margin, service burden. 🔴 not disclosed.
Unitree — low-cost access + developer workflow
- Proven: R1 around/from US$4,900 / US$5,900, G1 from US$13.5K, H2 US$29,900, plus H2 Plus / G1-D tooling anchors. 🟢 Unitree official pages; live checked 2026-06-17.
- Proof debt: shipments by model/customer, industrial uptime, repeat paid deployment, support/warranty burden, product-line gross margin, customer ROI. 🔴 not disclosed in reviewed pages.

Footer: Evidence map only. No winner ranking. No trade recommendation. Capacity ≠ production; deployment KPI ≠ economics; low price ≠ reliability/margin.

5. What would change the classification

Tesla upgrades toward S5 only if primary sources disclose multiple of:

Quarterly Optimus output, accepted units, or internally deployed units. 🟢
Fremont / Texas line yield, uptime, bottleneck, and cost-curve data. 🟢
Factory task success, runtime, intervention rate, reset rate, and safety data. 🟢
Internal factory productivity gain, cost saving, or ROI/payback. 🟢/🟡
External paid customer, contract value, robot count, pricing, repeat order, robot revenue, gross margin, warranty/service cost, or segment disclosure. 🟢

Figure upgrades toward S5 only if official or customer-side sources disclose multiple of:

Customer-confirmed ROI/payback or productivity improvement. 🟢/🟡
Robot count by customer/site and paid deployment terms. 🟢
Repeat orders across BMW / Catalyst / additional named customers. 🟢
Uptime/intervention distribution over long-duration deployments. 🟢/🟡
Revenue, gross margin, service cost, support burden, or fleet economics. 🟢/🟠

Unitree upgrades toward S5 only if official or customer-side sources disclose multiple of:

Humanoid shipments by model and customer type. 🟢
Industrial/commercial deployment runtime, uptime, intervention, and safety data. 🟢/🟡
Repeat orders, contract value, channel sell-through, service revenue, or enterprise customer references with economics. 🟢/🟡
Product-line gross margin, support cost, warranty/service burden. 🟢/🟠
Evidence Unitree captures software/data/deployment value rather than only hardware ASP. 🟢/🟠

6. Common misconceptions

Misconception 1: Tesla’s capacity number means current production leadership.

Correction: Tesla currently has the strongest public capacity-intent evidence, not public output/yield/deployment economics. 🟢/🟠

Misconception 2: Figure’s BMW data means humanoid commercialization is solved.

Correction: Figure’s BMW data is strong S4 evidence because it includes duration, shifts, runtime, task count, and production context; it still lacks ROI/payback, robot count, intervention rate, repeat order, revenue, margin, and service burden. 🟢/🟠

Misconception 3: Unitree’s low price proves winner status.

Correction: low price lowers experimentation cost and can pressure the industry cost curve; it does not prove industrial reliability, autonomy, support cost, gross margin, or customer ROI. 🟠

Misconception 4: A single leaderboard is the right frame.

Correction: the useful question is which bottleneck is becoming measurable first: capacity, deployment, cost, autonomy, service burden, or economics. 🟠

7. Think Deeper questions

Which first S5 evidence release matters most: Tesla accepted output/yield, Figure customer-verified repeat ROI, or Unitree verified shipment scale and margin?
If Unitree compresses humanoid ASP before autonomy is solved, who captures value: hardware OEM, model/data layer, integrator, service operator, or customer?
If Tesla keeps Optimus internal for years, should factory productivity/payback be treated as a substitute for robot revenue, or as a separate internal-efficiency thesis?
If Figure remains private but has the clearest deployment KPI, which public-market proxies can capture the signal without relying on supplier rumors?

8. Source list

Tesla:

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 🟢 original primary filing; live request returned HTTP 403 from this environment on 2026-06-17.
knowledge/robotics-tesla-optimus-sec-filing-excerpt-appendix-2026-06-16.md 🟠 local extraction / claim-check layer; source pages are 🟢 primary.

Figure:

Figure, “F.02 Contributed to the Production of 30,000 Cars at BMW”, 2025-11-19: https://www.figure.ai/news/production-at-bmw 🟢 HTTP 200 and key anchors matched 2026-06-17.
Figure, “Ramping Figure 03 Production”, 2026-04-29: https://www.figure.ai/news/ramping-figure-03-production 🟢 HTTP 200 and core anchors matched 2026-06-17; exact cadence phrase should be manually re-opened before quoting as 1 robot per hour.
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 🟢 for announcement; 🟠 for economics undisclosed; HTTP 200 and anchors matched 2026-06-17.

Unitree:

Unitree R1 product page: https://www.unitree.com/R1 🟢 HTTP 200 and price anchors matched 2026-06-17.
Unitree G1 product page: https://www.unitree.com/g1/ 🟢 HTTP 200 and price anchors matched 2026-06-17.
Unitree H2 product page: https://www.unitree.com/H2 🟢 HTTP 200 and price anchors matched 2026-06-17.
Unitree H2 Plus product page: https://www.unitree.com/H2plus 🟢 HTTP 200 and NVIDIA / Isaac / TeleOp / Sim anchors matched 2026-06-17.
Unitree G1-D product page: https://www.unitree.com/G1-D 🟢 HTTP 200 and data / model / G1-D anchors matched 2026-06-17.

9. Public-safe / internal-only boundary

Public-safe:

The evidence matrix, source grades, dated anchors, caveats, S3/S4/S5 classification, and slide/footer wording above.

Keep internal-only / exclude:

Hugo private portfolio data, watchlist sizing, trade rationale, private channel checks, paid-report excerpts, tax/legal context, buy/sell/hold language, and any supplier-customer inference not supported by primary or high-quality customer-side sources.