Run metadata:

• ran_at: 2026-08-11T03:15:00+03:00
• job_id: physical_ai_model_to_matter
• task: Task 1 — Physical AI / Embodiment Scan + Task 2 — Ethan Portfolio Engine
• scope: physical AI / embodiment / robotics / bionics / AI materials / BCI / defense robotics / capital-flow signals
• requires_web_current: TRUE
• output_intent: full report file + compact portfolio index

---

EXECUTIVE READ

The current physical-AI cycle is no longer only demo-driven. The strongest signals in this run are:

1. Platform layer acceleration: NVIDIA is moving aggressively from chips into full-stack physical-AI infrastructure — Cosmos 3, Isaac, GR00T, Jetson Thor, Halos safety, Omniverse/digital twins — with robotics, industrial, Japanese manufacturing, and humanoid partners adopting pieces of the stack. This strengthens NVDA as the clearest public “picks-and-shovels” exposure to embodied AI. Source URL: https://nvidianews.nvidia.com/news/nvid…REDACTED (nvidianews.nvidia.com)

2. Deployment proof is emerging but still narrow: Figure 02/03 at BMW Spartanburg is one of the cleanest public examples of humanoids moving from staged demos into a named manufacturing environment. BMW says Figure 03 is starting work on complex sequencing/logistics applications after prior Figure 02 deployment at Spartanburg. Source URL: https://www.press.bmwgroup.com/global/article/detail/T0458778EN/bmw-…REDACTED (press.bmwgroup.com)

3. Capital flow is now massive and strategic: NEURA announced a Series C of up to $1.4B; Apptronik announced a $520M Series A-X extension; Skild AI announced close to $1.4B in January 2026; these are not normal robotics rounds — they indicate a repricing of humanoid / robot-brain companies as strategic infrastructure assets. Source URLs: https://neura-robotics.com/record-series-c/ ; https://apptronik.com/news-collection/appt…REDACTED ; https://www.morningstar.com/news/business-wire/20260114335623/skil…REDACTED (neura-robotics.com)

4. Defense robotics has crossed into production/procurement signal: AeroVironment received a three-year $500M IDIQ for counter-UAS capabilities; Anduril and Shield AI both announced U.S. Air Force production-contract wins related to Collaborative Combat Aircraft mission autonomy. Source URLs: https://investor.avinc.com/news-releases/news-release-details/av-a…REDACTED ; https://www.anduril.com/news/u-s-…REDACTED ; https://shield.ai/shie…REDACTED/ (investor.avinc.com)

5. Bottleneck remains dexterity + safety + reliable deployment economics: recent tactile-sensing research shows progress in whole-arm sensing and adaptive manipulation, but the research literature still frames contact-rich manipulation as difficult and not yet broadly solved. Source URLs: https://www.nature.com/articles/s44460-026-00097-1 ; https://link.springer.com/article/10.1007/s44245-026-00275-y (nature.com)

6. AI materials / autonomous labs are becoming infrastructure, not just papers: NSF-backed cloud-programmable and autonomous lab networks are being funded at UChicago/Argonne, Johns Hopkins, CUNY, and CMU-linked programs, connecting robotics, AI agents, materials discovery, and national supply-chain strategy. Source URLs: https://www.computerscience.uchicago.edu/news/remo…REDACTED/ ; https://engineering.jhu.edu/news/buil…REDACTED/ ; https://asrc.gc.cuny.edu/headlines/2026/07/cuny…REDACTED/ ; https://www.cmu.edu/news/stories/archives/2026/july/conn…REDACTED (computerscience.uchicago.edu)

---

SIGNAL_LAYER

Physical-AI platform layer: NVIDIA is becoming the default embodied-AI substrate

Signal strength: HIGH.

NVIDIA launched Cosmos 3 on May 31, 2026 as an open world foundation model for physical AI, combining physical reasoning, world generation, and action prediction/generation; the company also announced a Cosmos Coalition with robotics and AI partners. Source URL: https://nvidianews.nvidia.com/news/nvid…REDACTED (nvidianews.nvidia.com)

NVIDIA’s March 2026 robotics announcement positioned Cosmos, Isaac simulation frameworks, Isaac GR00T models, Omniverse, Jetson, and industrial digital twins as a full robotics lifecycle platform, with listed ecosystem participants including ABB Robotics, FANUC, Figure, KUKA, Skild AI, Universal Robots, World Labs, and YASKAWA. Source URL: https://investor.nvidia.com/news/press-release-details/2026/NVID…REDACTED/ (investor.nvidia.com)

NVIDIA also announced Halos for Robotics in June 2026, framing safety as a full-stack system for robots and physical AI, with Agility Robotics named as the first humanoid/physical-AI partner incorporating elements into its safety architecture. Source URL: https://nvidianews.nvidia.com/news/nvid…REDACTED (nvidianews.nvidia.com)

Interpretation:
This is the cleanest public-market conversion path from “AI model intelligence” to “machines that act.” The investable signal is not that every robot wins; it is that simulation, synthetic data, edge inference, safety architecture, and digital twins become unavoidable toll roads.

Figure + BMW: real deployment signal, not just humanoid theater

Signal strength: HIGH, but still pilot-stage.

BMW announced on June 25, 2026 that Figure 03 would start at BMW Group Plant Spartanburg after a successful Figure 02 deployment, with Figure 03 working on complex sequencing applications in logistics. Source URL: https://www.press.bmwgroup.com/global/article/detail/T0458778EN/bmw-…REDACTED (press.bmwgroup.com)

Figure separately stated that Figure 03 arrived at BMW’s Hall 52 after Figure 02’s 2025 assembly-line deployment, and framed the new phase as Figure 03 + Helix 02 handling cognitive and dexterous manufacturing-logistics tasks. Source URL: https://www.figure.ai/news/f-03-at-bmw (figure.ai)

Figure’s November 2025 deployment writeup claimed its Figure 02 deployment at BMW ran for 11 months and involved sheet-metal loading in automotive manufacturing. Source URL: https://www.figure.ai/news/production-at-bmw (figure.ai)

Figure also announced in April 2026 that BotQ had produced over 350 Figure 03 units and increased production rate from 1 robot/day to 1 robot/hour. Source URL: https://www.figure.ai/news/ramp…REDACTED (figure.ai)

Interpretation:
This is one of the best deployment-quality signals in humanoids because it combines:

• named customer: BMW;
• named site: Spartanburg;
• manufacturing/logistics task context;
• robot generation transition: Figure 02 → Figure 03;
• manufacturing-scale claim: BotQ production ramp.

Caution:
This is not yet proof of broad humanoid ROI. The signal is “humanoid deployment is becoming operationally testable,” not “humanoids are already replacing industrial automation at scale.”

Tesla Optimus: capex/preparation signal, but execution risk remains high

Signal strength: MEDIUM.

Tesla’s Q1 2026 10-Q states that Tesla is making preparations and investments for large-scale production of Optimus. Source URL: https://www.sec.gov/Archives/edgar/data/1318605/000162828026026673/tsla-20260331.htm (sec.gov)

Tesla proxy material filed in April 2026 stated that first-generation production lines for Optimus were being installed in anticipation of volume production. Source URL: https://ir.tesla.com/_flysystem/s3/sec/000162828026026551/tsla-20260422-gen.pdf (ir.tesla.com)

The City of Fremont said in January 2026 that Tesla was retooling part of the Fremont campus for Optimus robot production and that Tesla had stated the retooling would not cause job losses and could increase Fremont headcount. Source URL: https://www.fremont.gov/Home/Components/News/News/1643/ (fremont.gov)

Interpretation:
Tesla’s signal is stronger on manufacturing preparation than on externally verified robot productivity. Tesla may have long-term leverage if it can combine vehicle-scale manufacturing, AI training infrastructure, actuation design, and internal factory deployment. But current evidence remains less deployment-verified than Figure/BMW.

Robot-brain / foundation model companies: private capital is buying the “portable cognition” layer

Signal strength: HIGH.

Skild AI announced close to $1.4B in January 2026 funding led by SoftBank Group, with participation from NVentures, Macquarie Capital, Bezos Expeditions, and others; Skild frames its “Skild Brain” as a general-purpose model across robot embodiments and tasks. Source URL: https://www.morningstar.com/news/business-wire/20260114335623/skil…REDACTED (morningstar.com)

Skild announced expanded collaborations with NVIDIA, ABB Robotics, Universal Robots, and MiR in March 2026, specifically to deploy its AI-powered robot brain across industries and applications. Source URL: https://www.globenewswire.com/news-release/2026/03/17/3256839/0/en/skil…REDACTED.html (globenewswire.com)

Google DeepMind’s Gemini Robotics-ER 1.6, released in April 2026, is described as a high-level reasoning model for robots that can call tools and make safer spatial decisions under constraints such as gripper limits or object weight. Source URL: https://deepmind.google/blog/gemi…REDACTED/ (deepmind.google)

Axios reported on July 30, 2026 that Google updated its robotics software with Gemini Robotics 2, with Google saying the model enables robots to reason through movement. Source URL: https://www.axios.com/2026/07/30/goog…REDACTED (axios.com)

Interpretation:
The “robot brain” layer may become the highest-value abstraction if it transfers across embodiments. But proof requires cross-hardware generalization, reliability, safety, and economic deployment.

Humanoid capital formation: acceleration and crowding

Signal strength: HIGH.

NEURA Robotics announced a Series C round of up to $1.4B in June 2026, with investors including Tether, Qualcomm Technologies, Amazon, NVIDIA, Bosch, Schaeffler, EIB, and others. Source URL: https://neura-robotics.com/record-series-c/ (neura-robotics.com)

Apptronik announced a $520M Series A-X in February 2026, with existing investors including B Capital, Google, Mercedes-Benz, and PEAK6, and new investors including AT&T Ventures, John Deere, and Qatar Investment Authority; Apptronik said the capital would ramp Apollo production and expand deployments. Source URL: https://apptronik.com/news-collection/appt…REDACTED (apptronik.com)

1X opened its NEO Factory in Hayward, California, with consumer shipments planned for 2026 and NVIDIA Jetson Thor / Isaac named as part of the stack. Source URL: https://www.globenewswire.com/news-release/2026/4/30/3285118/0/en/1x-o…REDACTED.html (globenewswire.com)

Interpretation:
Capital is chasing the same belief: humanoids are a general-purpose labor platform. The risk is valuation outrunning deployment. The key question for 2026–2027 is not demo capability; it is hours worked, failure rate, maintenance cost, and customer renewal.

Defense robotics / autonomy: procurement is real and accelerating

Signal strength: HIGH.

AeroVironment announced a three-year $500M IDIQ contract for counter-UAS/counter-small-UAS capabilities supporting JIATF-401’s Domestic Shield Program. Source URL: https://investor.avinc.com/news-releases/news-release-details/av-a…REDACTED (investor.avinc.com)

Anduril announced on June 17, 2026 that the U.S. Air Force selected it to bring mission autonomy to Collaborative Combat Aircraft and move mission autonomy toward production-ready capability. Source URL: https://www.anduril.com/news/u-s-…REDACTED (anduril.com)

Shield AI announced on June 17, 2026 that it was awarded a U.S. Air Force production contract for Collaborative Combat Aircraft mission autonomy. Source URL: https://shield.ai/shie…REDACTED/ (shield.ai)

Anduril also announced a May 2026 agreement to scale production of surface-launched Barracuda-500M to at least 1,000 all-up rounds per year, with first tranche deliveries expected in H1 2027; Anduril said Lattice integration can enable autonomous and collaborative behaviors. Source URL: https://www.anduril.com/news/andu…REDACTED (anduril.com)

Interpretation:
Defense robotics is currently more procurement-convertible than civilian humanoids. Counter-UAS, autonomous aircraft, loitering systems, and autonomy software are receiving contract dollars now.

Bionics / prosthetics / tactile intelligence: dexterity is the economic frontier

Signal strength: MEDIUM-HIGH for research, MEDIUM for immediate investability.

A July 2026 Nature Sensors paper presented EmArm, a robotic arm integrating rigid kinematics, large-area soft tactile skins, proprioception, and closed-loop perception-action control, with submillimetre tactile localization and contact-rich manipulation capabilities. Source URL: https://www.nature.com/articles/s44460-026-00097-1 (nature.com)

A 2026 Springer review states that dexterous robotic hands remain challenged by fragility, integration cost, tactile-sensing robustness, and control complexity under real contact dynamics. Source URL: https://link.springer.com/article/10.1007/s44245-026-00275-y (link.springer.com)

A Nature Communications paper on shared human-machine control of intelligent bionic hands found proof-of-concept benefits for transradial amputees, reducing cognitive burden while improving grasping. Source URL: https://www.nature.com/articles/s41467-025-65965-9 (nature.com)

Interpretation:
Walking is increasingly commoditized; useful hands remain hard. Tactile sensing, compliant actuation, shared control, sensor fusion, and durability are likely to determine which robots move from factory demos to economically valuable labor.

BCI / neural interfaces: clinically meaningful but not yet broad commercial revenue

Signal strength: MEDIUM.

ClinicalTrials.gov lists Neuralink’s PRIME study as a first-in-human early feasibility study evaluating the Neuralink N1 Implant and R1 Robot for people with tetraparesis or tetraplegia; the listing was last updated January 2026. Source URL: https://clinicaltrials.gov/study/NCT06429735 (clinicaltrials.gov)

Neuralink’s January 2026 “Two Years of Telepathy” update says Telepathy records neural activity from hand/arm motor regions and translates those signals into digital commands, bypassing damaged neural pathways. Source URL: https://neuralink.com/updates/two-years-of-telepathy/ (neuralink.com)

Synchron says its Stentrode is implanted via the jugular vein and is being assessed in clinical trials in the U.S. and Australia; Synchron also states its BCI system is investigational and not commercially approved in any geography. Source URL: https://synchron.com/ (synchron.com)

Interpretation:
BCI has strategic relevance to the human-machine boundary and future prosthetic/robotic control, but near-term portfolio conversion remains weak unless public-market exposure is via large medtech, semis, or private-market access. The highest-confidence claim is clinical progress; the lowest-confidence claim is near-term revenue scale.

AI materials / autonomous labs: Model → Matter infrastructure is being funded

Signal strength: HIGH for strategic infrastructure, MEDIUM for direct public investability.

UChicago/Argonne announced a $20M NSF grant for PoLARIS, a remotely operated robotic lab using AI, robotics, informatics, and standards for polymer/materials research across more than 20 autonomous stations. Source URL: https://www.computerscience.uchicago.edu/news/remo…REDACTED/ (computerscience.uchicago.edu)

Johns Hopkins announced a $20M NSF-backed AI-powered lab network focused on materials for extreme conditions and autonomous materials research. Source URL: https://engineering.jhu.edu/news/buil…REDACTED/ (engineering.jhu.edu)

CUNY announced an $18.1M NSF award to build AI-BIOME, a cloud-programmable national lab using AI and robotics for bio-inspired materials discovery, design, and production. Source URL: https://asrc.gc.cuny.edu/headlines/2026/07/cuny…REDACTED/ (asrc.gc.cuny.edu)

A 2026 Communications Materials article argues that multi-agent AI and autonomous labs can help navigate high-dimensional materials research, but also notes current technology gaps around IP protection and broader materials-research workflows. Source URL: https://www.nature.com/articles/s43246-026-01219-5 (nature.com)

Interpretation:
This is the “Model → Matter” bridge outside humanoids: AI systems designing, testing, and iterating physical materials using robotic labs. It matters for semiconductors, batteries, polymers, aerospace, defense, healthcare materials, and industrial supply chains.

CAPITAL_FLOW_MAP

Private capital

Public procurement / defense capital

Public-market pricing snapshot fetched this run

---

CONVERSION_LAYER

Highest conversion signals

1. Defense robotics → contracts → revenue - AVAV’s $500M IDIQ is a hard conversion signal. - Anduril/Shield AI CCA autonomy contracts show procurement appetite for autonomy-native defense companies. - Conversion path: threat environment → counter-UAS/autonomy budgets → contracts → production.

2. NVIDIA physical-AI stack → ecosystem adoption → compute/software demand - Cosmos, Isaac, GR00T, Jetson, Halos, Omniverse create a hardware/software loop. - Conversion path: robotics teams need simulation, synthetic data, edge inference, safety systems → NVIDIA stack adoption.

3. Figure/BMW → operational deployment → customer validation - BMW + Figure 03 is a named, credible deployment setting. - Conversion path: pilot → task expansion → uptime proof → procurement/revenue → supplier read-through.

4. Autonomous materials labs → grants/infrastructure → future industrial partnerships - NSF awards create institutional infrastructure; commercialization timing likely longer. - Conversion path: grants → shared lab networks → validated workflows → industry adoption.

Weak conversion signals

• Humanoid demo videos without customer/site/hour/task data.
• Preorders without evidence of shipped units or service economics.
• BCI milestones without regulatory approval or reimbursement path.
• AI-materials papers without reproducible lab execution or industrial buyer.

---

TIME_MAP

0–6 months: near-term monitoring window

• Track Figure 03 at BMW Spartanburg for actual task hours, uptime, number of deployed robots, and expansion beyond sequencing/logistics.
• Track Tesla Optimus for production-line status, internal deployment proof, and whether Tesla publishes robot-hours or real factory tasks.
• Track AVAV contract task orders and C-UAS revenue conversion.
• Track Anduril/Shield CCA program milestones, especially flight testing and production-readiness.
• Track NVIDIA adoption of Cosmos 3 Edge, Halos, GR00T, and Jetson Thor across named robotics partners.
• Track 1X consumer shipment reality vs factory/preorder narrative.

6–18 months: proof window

• Humanoid leaders must show:
• sustained paid deployment;
• repeatable tasks;
• lower cost per task/hour than alternatives;
• durable hands/actuators;
• low maintenance burden;
• customer renewal or expansion.
• Defense autonomy must show:
• scaled production;
• autonomy certification / operational acceptance;
• recurring procurement.
• Autonomous labs must show:
• reproducible discoveries;
• integration with industrial R&D;
• reduced cycle time vs traditional labs.

18–36 months: thesis validation window

• Physical AI becomes an investable mega-theme if robot intelligence + hardware reliability + deployment economics converge.
• If not, the sector stays infrastructure-heavy, with profits captured by chips, simulation, defense procurement, and industrial automation incumbents rather than humanoid OEMs.

---

BOTTLENECK_INDEX

---

THESIS_STATE

Current thesis

Physical AI is shifting from:

• demo phase → to infrastructure + deployment proof phase.

The strongest investable thesis is not “buy humanoids broadly.”
The stronger thesis is:

AI is moving into the physical world through a stack: compute → world models → simulation → robot foundation models → edge inference → safety → deployment → data flywheel.

Thesis upgraded

• NVIDIA / platform layer: upgraded due to breadth of Cosmos/Isaac/GR00T/Halos adoption.
• Defense robotics: upgraded due to concrete C-UAS and CCA production/procurement signals.
• Figure/BMW as deployment proof: upgraded from demo to credible pilot/early deployment signal.
• Autonomous labs / AI materials: upgraded as strategic infrastructure due to NSF-funded cloud-programmable lab networks.

Thesis unchanged

• Humanoid mass adoption: still unproven.
• Tesla Optimus: still high-upside but evidence remains more production-preparation than external deployment proof.
• BCI: clinically important but commercial/public-market conversion remains early.

Thesis downgraded / caution

• Any humanoid company without:
• named deployment;
• customer economics;
• uptime data;
• repeatable tasks;
• safety pathway;
• maintenance cost disclosure.

---

STRATEGIC INTERPRETATION

What matters most

1. Deployment beats demos.
   BMW/Figure and AVAV defense contracts are higher quality than viral humanoid videos.

2. The infrastructure layer may capture value before robot OEMs do.
   NVIDIA, cloud providers, edge semis, simulation software, safety systems, and industrial automation incumbents may monetize before humanoid makers prove unit economics.

3. Defense autonomy is the earliest high-conversion vertical.
   Defense customers pay for autonomy now because the pain is immediate: drone threats, contested airspace, force protection, and attritable systems.

4. Dexterity is still the choke point.
   Hands, tactile sensing, contact-rich manipulation, compliant actuation, and reliability are likely to decide winners.

5. Model → Matter is broader than humanoids.
   Autonomous labs and AI materials discovery are strategically important because they connect AI models to real-world experimental loops.

---

SOURCES

1. NVIDIA — “NVIDIA Launches Cosmos 3, the Open Frontier Foundation Model for Physical AI”
   URL: https://nvidianews.nvidia.com/news/nvid…REDACTED

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

3. NVIDIA — “NVIDIA Announces Halos for Robotics, the Industry’s First Full-Stack Safety System for Physical AI”
   URL: https://nvidianews.nvidia.com/news/nvid…REDACTED

4. BMW Group — “BMW Group advances the use of Physical AI in production with Figure 03 project in Spartanburg”
   URL: https://www.press.bmwgroup.com/global/article/detail/T0458778EN/bmw-…REDACTED

5. Figure AI — “F.03 Arrives at BMW”
   URL: https://www.figure.ai/news/f-03-at-bmw

6. Figure AI — “F.02 Contributed to the Production of 30,000 Cars at BMW”
   URL: https://www.figure.ai/news/production-at-bmw

7. Figure AI — “Ramping Figure 03 Production”
   URL: https://www.figure.ai/news/ramp…REDACTED

8. Tesla SEC filing — Q1 2026 Form 10-Q
   URL: https://www.sec.gov/Archives/edgar/data/1318605/000162828026026673/tsla-20260331.htm

9. Tesla Investor Relations filing — April 2026 proxy material
   URL: https://ir.tesla.com/_flysystem/s3/sec/000162828026026551/tsla-20260422-gen.pdf

10. City of Fremont — “Tesla’s Optimus Manufacturing in Fremont”
    URL: https://www.fremont.gov/Home/Components/News/News/1643/

11. Skild AI / Business Wire via Morningstar — “Skild AI Raises $1.4B, Now Valued Over $14B”
    URL: https://www.morningstar.com/news/business-wire/20260114335623/skil…REDACTED

12. Skild AI — “Skild AI Expands Generalized Robot Intelligence…”
    URL: https://www.globenewswire.com/news-release/2026/03/17/3256839/0/en/skil…REDACTED.html

13. Google DeepMind — “Gemini Robotics ER 1.6”
    URL: https://deepmind.google/blog/gemi…REDACTED/

14. Axios — “Google updates its AI software for robots”
    URL: https://www.axios.com/2026/07/30/goog…REDACTED

15. NEURA Robotics — “NEURA Robotics Announces Record Series C of up to $1.4B”
    URL: https://neura-robotics.com/record-series-c/

16. Apptronik — “Apptronik closes over $935M Series A”
    URL: https://apptronik.com/news-collection/appt…REDACTED

17. 1X Technologies / GlobeNewswire — “1X Opens NEO Factory in Hayward, CA”
    URL: https://www.globenewswire.com/news-release/2026/4/30/3285118/0/en/1x-o…REDACTED.html

18. AeroVironment — “AV Awarded $500 million IDIQ…”
    URL: https://investor.avinc.com/news-releases/news-release-details/av-a…REDACTED

19. Anduril — “U.S. Air Force Selects Anduril to Bring Mission Autonomy to Collaborative Combat Aircraft”
    URL: https://www.anduril.com/news/u-s-…REDACTED

20. Shield AI — “Shield AI awarded U.S. Air Force production contract…”
    URL: https://shield.ai/shie…REDACTED/

21. Anduril — “Anduril, Department of War Sign Production Agreement for Surface-Launched Barracuda-500M”
    URL: https://www.anduril.com/news/andu…REDACTED

22. Nature Sensors — “Embodied sensorimotor integration for whole-arm tactile sensing and adaptive robotic manipulation”
    URL: https://www.nature.com/articles/s44460-026-00097-1

23. Springer Nature — “A review of adaptive intelligence in tactile sensing robotic hands…”
    URL: https://link.springer.com/article/10.1007/s44245-026-00275-y

24. Nature Communications — “Shared human-machine control of an intelligent bionic hand…”
    URL: https://www.nature.com/articles/s41467-025-65965-9

25. ClinicalTrials.gov — Neuralink PRIME Study NCT06429735
    URL: https://clinicaltrials.gov/study/NCT06429735

26. Neuralink — “Two Years of Telepathy”
    URL: https://neuralink.com/updates/two-years-of-telepathy/

27. Synchron — official site / investigational-device statement
    URL: https://synchron.com/

28. UChicago / Argonne — PoLARIS NSF autonomous lab grant
    URL: https://www.computerscience.uchicago.edu/news/remo…REDACTED/

29. Johns Hopkins Engineering — AI-powered lab network for materials
    URL: https://engineering.jhu.edu/news/buil…REDACTED/

30. CUNY ASRC — AI-BIOME NSF award
    URL: https://asrc.gc.cuny.edu/headlines/2026/07/cuny…REDACTED/

31. Carnegie Mellon University — autonomous laboratories / AI agents / digital twins
    URL: https://www.cmu.edu/news/stories/archives/2026/july/conn…REDACTED

32. Communications Materials — “Managing autonomous materials labs with multi-agent AI…”
    URL: https://www.nature.com/articles/s43246-026-01219-5

33. Finance snapshot — NVDA, TSLA, AVAV, GOOGL, SYM fetched during this run
    URLs: finance tool snapshot cited inline above.