· investment-strategies · 3 min read
Noosphere Labs’ $10.25M: Madrona and Trilogy Back Stealth Physical AI
Ex-Meta AI research director Kevin Carlberg raised $10.25M for Noosphere Labs — human-centered physical intelligence — led by Trilogy Equity Partners with major Madrona participation.
Trilogy Equity Partners led a $10.25 million raise for Noosphere Labs, with major participation from Madrona, disclosed via SEC Form D and reported August 11, 2026. Founder/CEO Kevin Carlberg (ex–Meta Reality Labs / FAIR; Sandia; Stanford PhD) is still in stealth.
Key facts
| Field | Detail |
|---|---|
| Company | Noosphere Labs (Seattle area; stylized Noösphere) |
| Round | $10.25M (Form D) |
| Date | Filing / GeekWire coverage Aug 10–11, 2026 |
| Lead | Trilogy Equity Partners |
| Major participant | Madrona |
| Status | Stealth — product not publicly detailed |
| Thesis line | “Human-centered physical intelligence” |
| Founder | Kevin Carlberg; director Tyler Simpson (ex–Meta/Microsoft) |
Who would use the product — and for what job
Honest gap: Noosphere has not published a product, customers, or pricing.
Editorial inference from founder research (not company claims): Carlberg’s Meta work on assistive wearable agents (egocentric video/audio → intent) and Sandia real-time simulation points at users who need AI that understands what a person is doing in the physical world — smart glasses, mixed reality, robotics assist — where today’s models still fail practical usefulness tests (his NeurIPS benchmark found best models relevant only ~55% of the time).
Job hypothesis: help people act in the world (not only chat on screens) with agents that perceive context continuously.
Until launch, treat buyers as unknown.
Why now
- Physical AI / wearables capital is hot after multi-year Meta and Apple investment in glasses-class devices.
- Screen-native LLMs underperform on egocentric assistance — research gap = startup wedge.
- Pacific Northwest has dense Meta/Microsoft/robotics talent willing to leave for founder-led physical AI.
- Form D timing shows the raise closed even before public narrative — classic stealth seed pattern.
Why Trilogy + Madrona — portfolio fit
Trilogy (Bellevue) leading puts a local board close to the team. Madrona major participation adds Seattle multi-stage AI/software muscle (same week Madrona appears across other PNW AI stories).
Likely founder rationale: raise from investors who already diligence Northwest deep-tech talent and will not force a premature consumer launch narrative — keep stealth while hardware/ML mature.
| Dimension | Fit |
|---|---|
| Stage | Early / seed-scale Form D |
| Thesis | Physical / wearable intelligence |
| Geography | Seattle |
| Risk | Stealth execution; category definition |
Competitive map (category, not named product)
| Lane | Examples of pressure |
|---|---|
| Big Tech wearables AI | Meta, Apple, Google device stacks |
| Robotics foundation models | Physical AI labs and humanoid stacks |
| Screen agents | Chat/coding agents — different modality |
| Academic assistive-agent research | Benchmarks Carlberg co-authored |
When not to invent a pitch
- Wrong if you fabricate customers or a shipping product.
- Wrong if Form D dollars are treated as a priced Series A with disclosed valuation (valuation not public in GeekWire coverage).
- Wrong if “physical AI” is equated with humanoid robots without evidence.
Practical takeaway
- Founders: Stealth is fine when the Form D + pedigree carry the raise — but ship a crisp job statement before the next round.
- Investors: Underwrite founder research edge + PNW hiring, not slides.
- Scouts: Watch noospherelabs.ai and Carlberg’s publication trail for the first concrete ICP.
Sources
- GeekWire (Aug 11, 2026): https://www.geekwire.com/2026/ex-meta-ai-research-director-raises-10-25m-from-trilogy-and-madrona-for-stealth-physical-ai-startup/
- SEC Form D: https://www.sec.gov/Archives/edgar/data/2149216/000214921626000001/xslFormDX01/primary_doc.xml
- Related: /2026-august-11-12-investment-news-personal-ai-defense-health