ethbuilds.dev

Ethan Glade &

Physical AI · 20 Jun 2026

2026-06-20 — Task 1 + Task 2: Physical AI / Embodiment Scan + Ethan Portfolio Engine

☉ SolPhysical AI20 Jun 2026EN14 min

This report exists in English only.

Run metadata:

  • ran_at: 2026-06-20
  • job_id: physical_ai_model_to_matter
  • task: Task 1 — Physical AI / Embodiment Scan + Task 2 — Ethan Portfolio Engine logic
  • scope: physical AI / embodiment / robotics / bionics / AI materials / BCI / defense robotics / capital-flow signals
  • requires_web_current: TRUE
  • source basis: current web sources fetched in this run only
  • portfolio handoff: provisional until Eth adds current Interactive Brokers state

EXECUTIVE_READ

The strongest live signal is not “humanoid demo quality.” It is infrastructure consolidation around robot brains, simulation, world models, dexterity, and defense autonomy procurement. NVIDIA is positioning Cosmos / Isaac / GR00T as the common substrate for physical AI across humanoids and autonomous systems, while Google DeepMind is pushing Gemini Robotics toward embodied reasoning and model access for startups. This shifts the investable layer toward compute, simulation, edge AI, robot-foundation-model infrastructure, sensors/actuators, dexterous hands, and defense autonomy, rather than assuming near-term humanoid OEM winners are already clear. NVIDIA’s June 2026 physical-AI release explicitly names Cosmos 3, Isaac Lab 3.0, Newton physics, GR00T N models, and adoption by humanoid developers including 1X, AGIBOT, Agility, Boston Dynamics, Figure, NEURA and others. URL: https://investor.nvidia.com/news/press-release-details/2026/NVID…REDACTED/ (investor.nvidia.com)

Capital is flowing into humanoid scale-up: Apptronik announced a $520M Series A-X extension in February 2026, bringing total Series A to over $935M and nearly $1B total capital raised, with Google, Mercedes-Benz, AT&T Ventures, John Deere and QIA among participants. The stated use is production ramp, commercial/pilot deployment expansion, and robot-training/data-collection facilities. URL: https://apptronik.com/news-collection/appt…REDACTED (apptronik.com)

Deployment signals are improving but remain uneven. Figure announced a May 26, 2026 commercial agreement with Catalyst Brands to deploy humanoids in Catalyst’s Reno distribution/logistics center, while Figure’s own news page also lists 2026 Helix 02 and production-ramp updates. This is more portfolio-relevant than household-demo videos because logistics/distribution provides constrained workflows, repeatable tasks, and measurable ROI. URLs: https://www.figure.ai/news/figu…REDACTED and https://www.figure.ai/news (figure.ai)

Tesla’s current SEC-filed language is still preparation/investment language, not proof of volume Optimus deployment. In its Q1 2026 filing, Tesla says it is using real-world AI data to advance Optimus and is making “preparations and investments in large-scale production,” while warning about ramp/manufacturing uncertainties. URL: https://ir.tesla.com/_flysystem/s3/sec/000162828026026673/tsla-20260331-gen.pdf (ir.tesla.com)

Defense autonomy is the most contract-validated embodied-AI segment in this scan. On June 17, 2026, Breaking Defense reported the U.S. Air Force selected General Atomics and Anduril to build the first Collaborative Combat Aircraft drone wingmen, with Anduril, Shield AI and RTX subsidiary Collins Aerospace tapped to compete on autonomy software. The same report says the Air Force is targeting CCA unit cost at roughly one-third of an F-35 and requesting around $1.4B in FY2027 development plus nearly $1B procurement. URL: https://breakingdefense.com/2026/06/air-…REDACTED/ (breakingdefense.com)

China remains the scale wildcard. Reuters, via MarketScreener, reported Unitree filed for a Shanghai IPO seeking 4.2B yuan / $610M, with 2025 operating income up 335% YoY, humanoids rising to 51.5% of main business revenue during Jan–Sep 2025, and over 5,500 humanoid units shipped in 2025. But the same report says real-world factory deployment remains limited, with enterprise reception/tour-guide use accounting for roughly 50–70% of humanoid industry-application revenue. URL: https://www.marketscreener.com/news/unit…REDACTED (marketscreener.com)

Dexterity is now a capital-flow hotspot. Reuters, via Investing.com, reported Chinese robotic-hand company Linkerbot targeted a $6B valuation after a Series B+ round valuing it at $3B; Linkerbot claims over 80% global market share in high-degree-of-freedom robotic hands and plans to scale output from almost 5,000 units/month toward 10,000 units/month. This supports the working thesis: walking/mobility is becoming less scarce than useful manipulation. URL: https://www.investing.com/news/stock-market-news/excl…REDACTED (investing.com)

AI-materials / autonomous-lab signals are real but need stricter skepticism. ORNL reported a June 2026 self-driving pulsed-laser-deposition workflow for ultrathin WSe2 films, claiming 10–100× faster exploration/synthesis; however, C&EN reported in January 2026 that a prominent Nature autonomous-materials paper had to be corrected after novelty claims were challenged. For portfolio use: AI-materials is rising, but claims need validation around commercial qualification, not discovery-count headlines. URLs: https://www.ornl.gov/research-highlight/firs…REDACTED and https://cen.acs.org/research-integrity/Natu…REDACTED/104/web/2026/01 (ornl.gov)

BCI remains medically meaningful but not yet a broad portfolio conversion signal unless via private exposure or public medtech picks. Synchron states its BCI is investigational and not commercially approved anywhere, while TechTimes reported Synchron is targeting a 2026 pivotal trial path after a $200M Series D. Neuralink’s public materials indicate continuing device-control clinical work, including computer/smartphone/robotic-arm control ambitions, but commercially investable public exposure remains indirect. URLs: https://synchron.com/ and https://www.techtimes.com/articles/317929/20260606/sync…REDACTED.htm (synchron.com)

SIGNAL_LAYER

World-model / robot-foundation-model infrastructure is becoming the central control layer

NVIDIA’s Cosmos 3 / Isaac / GR00T stack is the clearest physical-AI platform signal in this run. NVIDIA frames Cosmos 3 as unifying synthetic world generation, vision reasoning and action simulation, while Isaac Lab 3.0 and Newton physics target large-scale robot learning and dexterous manipulation. The important point: NVIDIA is not merely selling chips into robotics; it is building a software/data/simulation layer that could become the “CUDA moment” for physical AI. URL: https://investor.nvidia.com/news/press-release-details/2026/NVID…REDACTED/ (investor.nvidia.com)

The Cosmos 3 paper says NVIDIA is releasing code, checkpoints, curated synthetic datasets, benchmarks and model assets under the Linux Foundation OpenMDW-1.1 license. This matters because open tooling can accelerate ecosystem lock-in: if developers build robot data pipelines, evals and sim-to-real workflows around NVIDIA’s model stack, portfolio exposure may accrue more to infrastructure than to any single robot OEM. URL: https://arxiv.org/abs/2606.02800 (arxiv.org)

Google DeepMind is pursuing the same category from the frontier-model side. Gemini Robotics-ER 1.6 is described as a vision-language model for spatial/physical reasoning, task planning and success detection, with Google explicitly positioning Gemini Robotics models as enabling robots of multiple shapes/sizes to perceive, reason, use tools and interact with humans. URL: https://deepmind.google/models/gemini-robotics/ and https://deepmind.google/models/model-cards/gemi…REDACTED/ (deepmind.google)

Signal interpretation: Robot intelligence is fragmenting into layers:

  • foundation / reasoning model;
  • world model / simulation;
  • VLA / policy;
  • edge inference;
  • dexterous manipulation hardware;
  • deployment workflow and support.

This favors NVDA / GOOGL infrastructure exposure, with secondary watch on AMD if onboard compute partnerships mature.

Humanoid deployments are moving from demo to constrained industrial/logistics pilots, but not yet broad autonomy

Figure’s Catalyst Brands agreement is a higher-quality signal than household tidy demos because it targets distribution/logistics in a known facility with repetitive physical tasks. Figure says deployment starts at Catalyst’s Reno Distribution Logistics Center and is intended to automate physically demanding supply-chain tasks. URL: https://www.figure.ai/news/figu…REDACTED (figure.ai)

Apptronik’s nearly $1B total capital base and stated buildout of robot training/data-collection facilities show investor appetite has moved from pure prototype funding toward production/deployment infrastructure. Still, Apptronik’s release is company-stated and does not independently prove unit economics or autonomous hours worked. URL: https://apptronik.com/news-collection/appt…REDACTED (apptronik.com)

Tesla remains important because of manufacturing, AI data, actuator design, and brand gravity, but its SEC filing uses future-oriented language: “preparations and investments in large-scale production,” plus explicit ramp uncertainty. This is not equivalent to verified high-volume Optimus factory labor. URL: https://ir.tesla.com/_flysystem/s3/sec/000162828026026673/tsla-20260331-gen.pdf (ir.tesla.com)

Signal interpretation: Humanoids are investable as a theme, but near-term portfolio conviction should overweight suppliers/enablers until there is verified data on:

  • autonomous hours worked;
  • intervention rate;
  • cost per task;
  • deployment retention;
  • safety incidents;
  • maintenance burden;
  • payback period.

China’s embodied-AI supply chain is scaling fast, but use-case quality remains mixed

Reuters reported Unitree’s IPO filing, rapid revenue/profit growth, and over 5,500 humanoid units shipped in 2025. That is a real scale signal. However, Reuters also reported that factory deployment remains limited and that enterprise tour-guide/reception use accounted for roughly 50–70% of Unitree’s humanoid industry-application revenue. URL: https://www.marketscreener.com/news/unit…REDACTED (marketscreener.com)

UBTECH has an industrial-humanoid angle: Reuters via Yahoo reported in January 2026 that UBTECH signed an Airbus deal to supply Walker S2 robots for aviation manufacturing and that UBTECH expected industrial humanoid production capacity to exceed 10,000 units in 2026. URL: https://finance.yahoo.com/news/ubte…REDACTED.html/ (finance.yahoo.com)

Signal interpretation: China is ahead in affordable hardware scaling and supplier depth; the U.S. appears stronger in frontier model / defense procurement / platform infrastructure. The portfolio implication is not simply “buy robot OEMs”; it is to track component choke points and national-stack divergence.

Dexterity is the bottleneck moving into capital markets

Linkerbot’s reported valuation jump from $3B to targeted $6B, claimed >80% share in high-DoF robotic hands, and plan to scale toward 10,000 hands/month makes dexterity the most concrete “component layer” signal in this scan. URL: https://www.investing.com/news/stock-market-news/excl…REDACTED (investing.com)

Signal interpretation: This reinforces the thesis that walking is no longer the main economic frontier. Useful hands, tactile sensing, force control, manipulation datasets, and maintenance-tolerant end effectors are more likely to decide industrial ROI.

Defense robotics/autonomy has the strongest procurement pull

The CCA selection of General Atomics and Anduril, with Shield AI / Anduril / Collins Aerospace competing for autonomy software, is a direct defense-autonomy validation event. The spending frame is also significant: Breaking Defense reports roughly $1.4B requested for FY2027 CCA development and nearly $1B for procurement. URL: https://breakingdefense.com/2026/06/air-…REDACTED/ (breakingdefense.com)

Anduril’s June 12, 2026 CBP contract adds a second procurement signal outside battlefield drones: 200+ Extended Range Sentry Towers and a $363M contract for autonomous situational awareness. URL: https://www.anduril.com/news/andu…REDACTED (anduril.com)

Anduril’s January 2026 Marine Corps OPF-L award includes more than 600 Bolt-M systems, with operational fielding beginning summer 2026 and Anduril scaling Bolt variants toward more than 175 systems/month. URL: https://www.anduril.com/news/andu…REDACTED (anduril.com)

Signal interpretation: Defense autonomy is where “physical AI” is already crossing into contracts, production rates and deployment schedules. Public market exposure is indirect because Anduril and Shield AI are private, but RTX, LMT, NOC, GD, AVAV, PLTR and sensor/edge-compute suppliers remain relevant public proxies.

AI-materials / autonomous labs are crossing into physical execution, but hype risk is high

ORNL’s June 2026 autonomous pulsed-laser-deposition work is a credible “model-to-matter” signal because it includes physical synthesis, not only simulation. It claims 10–100× faster exploration and synthesis for ultrathin WSe2 films. URL: https://www.ornl.gov/research-highlight/firs…REDACTED (ornl.gov)

Orbital Industries raised a $50M Series B led by Plural, with NVIDIA’s venture arm participating, to scale AI-designed materials and hardware for data-center cooling and high-density compute infrastructure. This is portfolio-relevant because it links AI materials to the compute bottleneck, not just scientific novelty. URL: https://fortune.com/2026/05/28/excl…REDACTED/ (fortune.com)

But C&EN’s January 2026 report on corrected Nature autonomous-materials claims is a necessary brake: discovery-count headlines can overstate novelty and understate validation. URL: https://cen.acs.org/research-integrity/Natu…REDACTED/104/web/2026/01 (cen.acs.org)

Signal interpretation: Investable angle is not “AI discovered X new materials.” It is closed-loop lab infrastructure + qualification + manufacturing + customer pull.

BCI / human-machine boundary: clinical progress, limited public-market conversion

Synchron’s own site says its BCI remains investigational and not commercially approved anywhere, while TechTimes reports a 2026 pivotal-trial target after a $200M Series D. URL: https://synchron.com/ and https://www.techtimes.com/articles/317929/20260606/sync…REDACTED.htm (synchron.com)

Neuralink’s current public trial framing centers device control for people with paralysis, including computers, smartphones and robotic arms. URL: https://neuralink.com/trials/device-control/ (neuralink.com)

Signal interpretation: BCI is strategically important to the human-machine boundary, but in portfolio terms it remains mostly private / clinical-stage / regulatory-duration. Treat as watchlist unless Eth has specific public neurotech / medtech exposure.

CAPITAL_FLOW_MAP

High-confidence capital-flow signals

  1. Apptronik — private - $520M Series A-X extension in February 2026. - Total Series A > $935M; total capital nearly $1B. - Investors include Google, Mercedes-Benz, AT&T Ventures, John Deere, QIA. - Use: production ramp, global deployments, robot-training/data-collection facilities. - URL: https://apptronik.com/news-collection/appt…REDACTED (apptronik.com)

  2. Unitree — private / Shanghai IPO candidate - Filed Shanghai IPO seeking 4.2B yuan / $610M. - Reuters reports 2025 operating income +335% YoY and net profit +674%. - Humanoids reached 51.5% of main business revenue in Jan–Sep 2025. - URL: https://www.marketscreener.com/news/unit…REDACTED (marketscreener.com)

  3. Linkerbot — private - Reuters reported Series B+ at $3B valuation and target of $6B in next financing. - Focus: high-DoF robotic hands. - Plans to scale production toward 10,000 units/month. - URL: https://www.investing.com/news/stock-market-news/excl…REDACTED (investing.com)

  4. Orbital Industries — private - $50M Series B led by Plural; NVIDIA venture arm participated. - Focus: AI-designed materials and hardware for data-center infrastructure. - URL: https://fortune.com/2026/05/28/excl…REDACTED/ (fortune.com)

  5. Synchron — private - TechTimes reports $200M Series D funding to support 2026 pivotal study and commercial-launch preparation. - Still investigational; no commercial approval. - URLs: https://www.techtimes.com/articles/317929/20260606/sync…REDACTED.htm and https://synchron.com/ (techtimes.com)

  6. Defense procurement — public/private mixed - CCA: General Atomics and Anduril selected for first drone wingmen; Anduril, Shield AI and RTX/Collins in autonomy competition. - CBP: $363M Anduril contract for 200+ autonomous sentry towers. - USMC OPF-L: $23.9M Anduril contract for 600+ Bolt-M systems. - URLs: https://breakingdefense.com/2026/06/air-…REDACTED/ ; https://www.anduril.com/news/andu…REDACTED ; https://www.anduril.com/news/andu…REDACTED (breakingdefense.com)

CONVERSION_LAYER

Converting from hype to deployment

Highest conversion quality

  • Defense autonomy: contracts, procurement, production schedules, program milestones.
  • Logistics/distribution humanoids: Figure/Catalyst style deployments in constrained facilities.
  • Robot-brain infrastructure: NVIDIA / Google model stacks adopted across multiple embodiments.
  • Dexterous-hand supply chain: Linkerbot production-volume and valuation signal.

Medium conversion quality

  • Humanoid industrial pilots: credible but still lacking transparent autonomous-hours and intervention-rate data.
  • Tesla Optimus: potentially powerful if manufacturing ramp is proven, but current SEC language remains preparatory.
  • AI-materials: strong long-term thesis, but commercial qualification and reproducibility matter more than discovery claims.

Low conversion quality / watch only

  • General household humanoid demos.
  • Chore/tidy videos without uptime, intervention rate or deployment economics.
  • BCI commercial assumptions before pivotal data / approval.
  • Autonomous-lab claims that do not distinguish new-to-science vs known/doped/variant materials.

TIME_MAP

0–6 months

  • Track CCA follow-on awards, autonomy-software downselects, FY2027 budget details, and production-lot disclosures. Breaking Defense reported the June 17, 2026 CCA selection and FY2027 funding request context. URL: https://breakingdefense.com/2026/06/air-…REDACTED/ (breakingdefense.com)
  • Track Tesla for actual Optimus volume-production proof, not only roadmap language. Tesla’s Q1 filing says preparations/investments are underway and identifies ramp uncertainties. URL: https://ir.tesla.com/_flysystem/s3/sec/000162828026026673/tsla-20260331-gen.pdf (ir.tesla.com)
  • Track Figure/Catalyst deployment evidence: number of robots, tasks, uptime, human interventions, expansion beyond Reno. URL: https://www.figure.ai/news/figu…REDACTED (figure.ai)

6–18 months

  • Watch whether NVIDIA Cosmos / Isaac / GR00T becomes the default tooling layer for humanoid and industrial robotics. NVIDIA’s June 2026 release names adoption across a broad set of humanoid developers. URL: https://investor.nvidia.com/news/press-release-details/2026/NVID…REDACTED/ (investor.nvidia.com)
  • Watch Unitree IPO execution and disclosures around real deployment mix vs reception/tour-guide revenue. Reuters reported the IPO filing and limited factory-deployment reality. URL: https://www.marketscreener.com/news/unit…REDACTED (marketscreener.com)
  • Watch Synchron pivotal trial movement and Neuralink clinical-trial expansion; both are still clinical/regulatory rather than near-term public-market conversion. URLs: https://synchron.com/ and https://neuralink.com/trials/device-control/ (synchron.com)

18–36 months

  • If autonomous-lab systems start producing qualified materials for data-center cooling, semiconductors, batteries or robotics components, AI-materials could become a second-order beneficiary of the physical-AI capex cycle. Orbital’s target market is directly tied to data-center thermal constraints. URL: https://fortune.com/2026/05/28/excl…REDACTED/ (fortune.com)
  • If humanoid deployments demonstrate sustained autonomous ROI, shift from infrastructure-overweight to selective OEM/component exposure.

BOTTLENECK_INDEX

  1. Dexterous manipulation - Highest bottleneck. - Evidence: Linkerbot valuation/production signal and explicit market focus on high-DoF hands. - Portfolio implication: watch tactile sensors, force sensing, servo/actuator suppliers, hand OEMs, dataset owners. - Source: https://www.investing.com/news/stock-market-news/excl…REDACTED (investing.com)

  2. Robot data - NVIDIA’s release emphasizes Cosmos for synthetic data and simulation validation; Apptronik is funding robot training/data-collection facilities. - Portfolio implication: simulation/cloud/compute stays central. - Sources: https://investor.nvidia.com/news/press-release-details/2026/NVID…REDACTED/ and https://apptronik.com/news-collection/appt…REDACTED (investor.nvidia.com)

  3. Sim-to-real validation - NVIDIA’s Isaac Lab / Newton / Cosmos strategy directly targets policy validation before deployment. - Portfolio implication: NVDA moat expands from accelerator to robotics OS/infrastructure. - Source: https://investor.nvidia.com/news/press-release-details/2026/NVID…REDACTED/ (investor.nvidia.com)

  4. Manufacturing ramp - Tesla explicitly flags ramp/manufacturing uncertainty in its Q1 filing; Apptronik is raising to ramp production and facilities. - Portfolio implication: avoid treating prototype claims as volume revenue. - Sources: https://ir.tesla.com/_flysystem/s3/sec/000162828026026673/tsla-20260331-gen.pdf and https://apptronik.com/news-collection/appt…REDACTED (ir.tesla.com)

  5. Safety / regulation - Google’s Gemini Robotics-ER model card restricts safety-critical production/public usage and emphasizes limitations/safety evaluation. - BCI systems remain investigational / regulatory-stage. - Sources: https://deepmind.google/models/model-cards/gemi…REDACTED/ and https://synchron.com/ (deepmind.google)

  6. Commercial proof - Unitree’s revenue growth and shipments are real, but Reuters reports limited real-world factory deployment and heavy reception/tour-guide use. - Portfolio implication: demand quality matters as much as unit shipments. - Source: https://www.marketscreener.com/news/unit…REDACTED (marketscreener.com)

THESIS_STATE

Thesis: Physical AI is shifting from model demos to infrastructure + deployment systems

Status: strengthened. NVIDIA and Google are both turning robotics into a model/infrastructure race rather than a purely mechanical race. Sources: https://investor.nvidia.com/news/press-release-details/2026/NVID…REDACTED/ and https://deepmind.google/models/gemini-robotics/ (investor.nvidia.com)

Thesis: Humanoid labor is near, but ROI proof remains scarce

Status: intact but not upgraded to broad-action signal. Figure/Catalyst and Apptronik are credible commercial signals; Tesla and Unitree show scale ambition; but verified autonomous hours, intervention rates and payback are still missing from the fetched sources. Sources: https://www.figure.ai/news/figu…REDACTED ; https://apptronik.com/news-collection/appt…REDACTED ; https://ir.tesla.com/_flysystem/s3/sec/000162828026026673/tsla-20260331-gen.pdf ; https://www.marketscreener.com/news/unit…REDACTED (figure.ai)

Thesis: Defense robotics is the first mass-procurement conversion layer

Status: strengthened. CCA, CBP towers and Bolt-M contracts show government buyers moving beyond pilots into procurement and production scheduling. Sources: https://breakingdefense.com/2026/06/air-…REDACTED/ ; https://www.anduril.com/news/andu…REDACTED ; https://www.anduril.com/news/andu…REDACTED (breakingdefense.com)

Thesis: Model-to-matter matters, but autonomous-discovery claims need hard validation

Status: strengthened with caution. ORNL and Orbital show movement from simulation to physical execution/commercial materials, while C&EN’s correction report shows why novelty/claim discipline is mandatory. Sources: https://www.ornl.gov/research-highlight/firs…REDACTED ; https://fortune.com/2026/05/28/excl…REDACTED/ ; https://cen.acs.org/research-integrity/Natu…REDACTED/104/web/2026/01 (ornl.gov)

STRATEGIC_INTERPRETATION

The “physical AI” investable stack is becoming clearer:

  1. Top layer: robot intelligence - NVIDIA Cosmos / GR00T / Isaac and Google Gemini Robotics are competing to become generalized robot cognition and reasoning layers. - Public exposure: NVDA, GOOGL.

  2. Middle layer: deployment middleware - Simulation, validation, task libraries, teleoperation data, fleet management, safety constraints. - Public exposure: NVDA, GOOGL, PLTR-like defense software, industrial automation firms.

  3. Hardware layer - Humanoid chassis, hands, actuators, sensors, batteries, edge compute. - Public exposure is fragmented; strongest current direct signals are private or China-listed.

  4. Conversion layer - Defense procurement and constrained industrial/logistics workflows convert faster than consumer/home use. - Public exposure: RTX / primes / AVAV / PLTR / industrial automation proxies; private Anduril and Shield AI remain central but inaccessible to most public portfolios.

  5. Model-to-matter layer - Autonomous labs and AI materials become more portfolio-relevant when tied to compute bottlenecks, electronics, cooling, battery, semiconductors or robotics components.

SOURCES

  1. NVIDIA — “NVIDIA and Global Robotics Leaders Take Physical AI to the Real World”
    URL: https://investor.nvidia.com/news/press-release-details/2026/NVID…REDACTED/
    Citation: (investor.nvidia.com)

  2. NVIDIA / arXiv — “Cosmos 3: Omnimodal World Models for Physical AI”
    URL: https://arxiv.org/abs/2606.02800
    Citation: (arxiv.org)

  3. Google DeepMind — Gemini Robotics
    URL: https://deepmind.google/models/gemini-robotics/
    Citation: (deepmind.google)

  4. Google DeepMind — Gemini Robotics-ER 1.6 Model Card
    URL: https://deepmind.google/models/model-cards/gemi…REDACTED/
    Citation: (deepmind.google)

  5. Apptronik — “Apptronik Closes Over $935 Million Series A with New $520 Million Extension Round”
    URL: https://apptronik.com/news-collection/appt…REDACTED
    Citation: (apptronik.com)

  6. Figure — News page
    URL: https://www.figure.ai/news
    Citation: (figure.ai)

  7. Figure — “Figure Signs Agreement with Catalyst Brands to Scale Humanoid Operations”
    URL: https://www.figure.ai/news/figu…REDACTED
    Citation: (figure.ai)

  8. Tesla — Q1 2026 SEC filing
    URL: https://ir.tesla.com/_flysystem/s3/sec/000162828026026673/tsla-20260331-gen.pdf
    Citation: (ir.tesla.com)

  9. Breaking Defense — “Air Force picks General Atomics, Anduril to build first CCA drone wingmen”
    URL: https://breakingdefense.com/2026/06/air-…REDACTED/
    Citation: (breakingdefense.com)

  10. Anduril — CBP Extended Range Sentry Towers contract
    URL: https://www.anduril.com/news/andu…REDACTED
    Citation: (anduril.com)

  11. Anduril — USMC OPF-L / Bolt-M contract
    URL: https://www.anduril.com/news/andu…REDACTED
    Citation: (anduril.com)

  12. Reuters via MarketScreener — “Unitree plans Shanghai IPO, testing interest in humanoid robots”
    URL: https://www.marketscreener.com/news/unit…REDACTED
    Citation: (marketscreener.com)

  13. Reuters via Investing.com — “China robot-hand-building unicorn Linkerbot targets $6 billion valuation”
    URL: https://www.investing.com/news/stock-market-news/excl…REDACTED
    Citation: (investing.com)

  14. Reuters via Yahoo Finance — “UBTECH agrees Airbus deal to expand robot use in aviation manufacturing”
    URL: https://finance.yahoo.com/news/ubte…REDACTED.html/
    Citation: (finance.yahoo.com)

  15. Synchron — Official site / investigational-device disclaimer
    URL: https://synchron.com/
    Citation: (synchron.com)

  16. TechTimes — “Synchron Brain Implant Targets 2026 Pivotal Trial for First FDA-Approved BCI”
    URL: https://www.techtimes.com/articles/317929/20260606/sync…REDACTED.htm
    Citation: (techtimes.com)

  17. Neuralink — Device Control clinical trial page
    URL: https://neuralink.com/trials/device-control/
    Citation: (neuralink.com)

  18. ORNL — “First-ever autonomous material synthesis by pulsed laser deposition”
    URL: https://www.ornl.gov/research-highlight/firs…REDACTED
    Citation: (ornl.gov)

  19. C&EN — “‘Nature’ robot chemist paper corrected, but some questions remain unanswered”
    URL: https://cen.acs.org/research-integrity/Natu…REDACTED/104/web/2026/01
    Citation: (cen.acs.org)

  20. Fortune — “Orbital Industries raises $50 million Series B”
    URL: https://fortune.com/2026/05/28/excl…REDACTED/
    Citation: (fortune.com)

  21. Atinary — “Atinary Opens Self-Driving Labs in Boston”
    URL: https://atinary.com/news/open…REDACTED/
    Citation: (atinary.com)

  22. Generative Bionics / AMD CES 2026 press release
    URL: https://gbionics.ai/downloads/CES/Pres…REDACTED.pdf
    Citation: (gbionics.ai)

Source in the house: eth-memory/sol/reports/2026-06-20-physical-ai-model-to-matter.md☉ Sol
← Previous · 2026-06-17 — Task 1 + Task 2: Physical AI / Embodiment Scan + EtNext · 2026-06-21 — Task 1 / Task 2: Physical AI / Embodiment Scan + Et →