Uber Beat Wayve Beat Waymo.

Good morning, crowd. 

It is finally Friday!

Robotics edition.

• Humanoids are coming.

• Robotaxis are here.

• The race for the robot brain just got real.

Here's who's winning the week.

Read time: 4 mins

The Physical AI Boom Reaches The Farm

Most AI lives on a screen. This kind drives itself through a soybean field at 2 a.m. and cuts weeds to the centimeter, with no herbicide. has 82 machines running across 16 states. This year's fleet sold out, and every robot was delivered. 

The platform keeps adding jobs: weeding now, feeding, spraying, and cover-crop planting on the way. Farmers are done with chemicals linked to Parkinson's and tied to the price of oil. 

The pull is real, and the machines work. lets everyday investors own a stake ahead of the scale-up the company is building toward.

^{Greenfield Robotics is Testing The Waters under tier 2 of Regulation A. No money or other consideration is being solicited, and if sent in response will not be accepted. No offer to buy the securities can be accepted and no part of the purchase price can be received until the offering statement filed by the company with the SEC has been qualified by the SEC. Any such offer may be withdrawn or revoked, without obligation or commitment of any kind, at any time before notice of acceptance given after the date of qualification. An indication of interest involves no obligation or commitment of any kind. “Reserving” shares is simply an indication of interest. There is no binding commitment for investors that reserve shares in this manner to ultimately invest and purchase the shares reserved of the company, or to purchase any shares of the company whatsoever.}

FRESH NEWS

Winner: Wayve (Speed + Generalization) Loser: Waymo (Infrastructure + Mapped Precision)

The Pattern: Uber and Wayve just launched robotaxis in London with cameras and radar, no lidar, no HD maps, just AI that generalizes to new streets. Waymo is still preparing its London launch with the traditional stack. This isn't a tie. Speed to market matters more than sensor perfection.

Why Wayve wins:

• Launched first (Uber partnership = distribution)

• Lighter tech stack (cheaper, scales faster)

• Generalization works (AI learns streets without mapping)

• Waymo's lidar moat is collapsing (proven redundant)

• Uber chose Wayve over investing in its own tech (capital efficiency signal)

Why Waymo's in trouble:

• Delayed London launch (second-mover disadvantage)

• Heavy infrastructure (lidar + HD maps = capital intensive)

• Waymo owns the tech but not the distribution (doesn't own Uber)

• Regulators are getting comfortable with camera-only (Wayve proved it works)

What founders should see: The robotaxi winner isn't the best self-driving tech—it's the best tech + distribution combination. Waymo has the tech. Uber has the distribution. Wayve + Uber have both. Capital efficiency wins wars. Waymo's approach costs 3-5x more per car.

What investors should see: If Wayve succeeds with cameras only, the competitive moat for lidar companies (Luminar, Ouster) shrinks dramatically. Wayve just became worth $2B+ (if valued like a robotaxi operator, not a tech vendor).

Images 2.0

Winner: OpenAI (Brain + Distribution) Loser: Figure (Now Competing With Its Former Supplier)

The Pattern: Sam Altman just said OpenAI "will definitely" build humanoids. They're hiring for actuators, mechanical design, simulation. Meanwhile, Figure—which dropped OpenAI in 2025 to build its own robot AI—is now competing against OpenAI for both talent and customer relationships.

Why OpenAI wins:

• Full-stack integration (brain + body, not just software)

• Training data flywheel (builds robots → collects training data → better models → better robots)

• Capital to move fast ($10B+ burning rate, can fund robotics indefinitely)

• Humanoids make sense (world designed for human form factor)

• Form factor doesn't matter if brain is 10x better (Altman's insight)

Why Figure's in trouble:

• Lost its AI supplier (now competing with OpenAI)

• Needs to rebuild trust as robot maker (not just AI importer)

• Competing on margins, not technology (OpenAI can out-spend them)

• Humanoid market might not be worth the hardware burn (see Tesla's bot struggles)

What founders should see: OpenAI treating robotics as an AI data generation problem, not a robotics problem. They're building robots to collect training data for their models. Figure's trying to sell robots. OpenAI's trying to build AGI. Different game. Figure loses.

What investors should see: If OpenAI's humanoid succeeds, it's a $100B+ business (robot + AI = vertical integration). If it fails, it's just another billion burned on hardware. Bet on the brain, not the body.

MIT

Winner: Motional/MIT (Interpretability) Loser: Pure Speed Players (Waymo, Wayve, Everyone Chasing Performance)

The Pattern: MIT built CW-Net, which translates a self-driving car's reasoning into English in real time. "Approaching stopped vehicle." "Close to cyclist." Suddenly, you can watch a robotaxi's brain working and predict failures before they happen.

Why interpretability wins:

• Regulators will demand it (they don't trust black boxes)

• Insurance companies will require it (liability reduction)

• Safety drivers will catch failures early (reduces accident rates)

• Debugging becomes possible (instead of retroactive crash analysis)

• Trust builds faster (humans can see the reasoning)

Why pure speed loses:

• Who cares if you hit 95% accuracy if regulators won't approve?

• Insurance costs eat all the margin (without trust)

• One accident = months of regulatory setback

• Interpretability is now table stakes, not a differentiator

What founders should see: If you're building self-driving, you can't just chase accuracy numbers. You need interpretability (CW-Net approach), or you'll get blocked by regulators and insurance. This is a 2-year head start for whoever builds this into their product.

What investors should see: Motional just got a moat (interpretable robotaxis = regulatory advantage). This is worth $500M-$1B in competitive advantage because it collapses regulatory risk.

Skild Ai

Winner: Skild (Generalization). Loser: Every Robot Company Still Fine-Tuning Per Task

The Pattern: Skild's S1 watches a human do a task once (11 minutes from demo to autonomous execution), then runs it. No fine-tuning. No new training data. Just one video. Generalist (competitor) just did the same thing. The game changed this week.

Why generalization wins:

• 380 hours of robot training → 11 minutes of human demo

• Tasks don't need to be in pretraining (S1 learned potting plants, flipping pancakes)

• Scaling becomes possible (one video per new task vs. 100s of hours)

• Deployment time collapses (11 minutes vs. weeks)

• Capital efficiency explodes (no custom data collection per client)

Why fine-tuning loses:

• Labor intensive (robot teleop takes 50-100 hours per task)

• Capital heavy (custom training data collection)

• Slow deployment (weeks of tuning before production)

• Company-specific (can't port skills to new robots)

What founders should see: If you're building anything that requires robot learning, buy Skild's model, don't build your own. The foundation model era is here. Competing on model training is like competing on building your own GPUs—you've already lost.

What investors should see: Skild just made every other robot company's training infrastructure obsolete. $14B valuation is defensible if this scales. Robot foundation models (Skild, Generalist, others) are the winners. Robot companies using them are the customers.

🚀 QUICK HITS

🔐 Gemini Goes Everywhere — Google confirms Gemini will eventually replace Assistant on Google Home, Google TV, and Android Automotive (dates TBD). Full ecosystem transition coming. Voice search is about to get messier (or smarter).

📊 AI Max Testing Expands — Google's AI-powered ad campaigns now allow A/B testing across multiple Search campaigns. Advertisers can test different budgets and ROI targets simultaneously. Automation eating media buying.

💼 Anthropic's Claude Context Explodes — Claude Sonnet 4 now handles 1M token windows (750K words, 75K lines of code in one prompt). Your entire codebase fits in one message. Code understanding shifts overnight.

🏭 Humanoid Robots Ramping Manufacturing — Tesla, Boston Dynamics, and Chinese competitors scaling from prototypes to actual factory floors. Physical AI is no longer theoretical—factories are hiring robots.

💳 Stripe's AI Infrastructure Play Continues — After the $7.5B OpenRouter acquisition, Stripe is now the de facto infrastructure bank for AI companies. Every AI startup needs payment processing; Stripe owns the pipeline.

🌐 Crypto Regulation Softening Post-Election — Treasury announcement of dollar buybacks signals soft stance on crypto. Institutional money sees opening. First time since 2021 that regulation isn't the headwind.

📱 Apple Foldable Still a Week Away — September 9 launch for iPhone Fold at $2,500. Supply chain stress building (component shortages). Samsung losing sleep.

🔬 Open-Source AI Quality Hitting 95% of Proprietary — Meta's Llama 4 now competitive with OpenAI and Anthropic on reasoning tasks. Commercial moat collapsing.

💰 Series A Funding Hitting All-Time Peaks — Companies raising in Sept-Oct getting peak valuations (but also peak competition). If you're Series A hunting, this is the expensive window.

🌍 Asia Capital Re-Entering U.S. Private Markets — Chinese, Japanese, Korean VCs deploying $40B+ into Silicon Valley. Geopolitical capital flows reshaping startup ecosystem.

🚗 TESLA CYBERCAB, WAYMO EXPANSION, WANDERCRAFT HUMANOID — THE RACE IS HEATING UP

Stories consolidating the winner/loser thesis:

Tesla Cybercab (Purpose-built, no steering wheel) — Founder play. Elon building end-to-end (hardware + software + manufacturing). High-risk, high-upside. If it works, Tesla owns robotaxis + manufacturing. If it fails, it's a Roadster-sized bet.

Waymo Expansion (Denver, San Diego, Tampa paid driverless) — Waymo's playing defense. Expanding paid service before London launch (after Wayve beats them). Capital-heavy, time-consuming. By the time Waymo reaches UK scale, Wayve + others have 2-year head start.

Wandercraft Humanoid (€100M at €750M valuation) — European play, raising euros in euros. Smart (avoids currency risk). But humanoid robotics is capital intensive. If OpenAI or Tesla kills it, Wandercraft becomes a footnote. Timing risk is existential.

Lyte Robotics ($165M at $1.6B) — Custom chips + perception systems. Smart vertical integration. But perception might not be the moat (Wayve proved cameras work). Chip play is real, but chip + hardware combo is risky.

🚀 QUICK HITS: WHO WINS THE WEEK

🚀 Waymo Debt Financing ($3B+) — WINNER: Shows confidence + capital efficiency (debt, not dilution). LOSER: Signals need for capital (cash burn is real). Who funds driverless robotaxis? Pimco + Blackstone (capital markets finally believe).

🎬 GoPro → Robotics Optics (Starman $285M acquisition) — QUESTION MARK: GoPro pivoting to robotics optics is smart (cameras for robots). But GoPro's brand is action cameras, not enterprise robotics. Starman betting on founder execution (ex-Gopro team). High risk.

🤖 Hugging Face Microduck ($399, 10K+ sold) — CONSUMER ROBOTICS WINNER: Sub-$400 robot selling like hardware. This is the signal that consumer robotics is here. Robot arms next. LOSER: Every robot company selling $50K+ industrial bots (consumer robotics is the real market).

✋ X Square Robot TwinDEX (Wearable + Robot Hand) — TRAINING DATA WINNER: Collecting dexterous training data from humans using exoskeletons + robot hands is genius (real-world + teleoperated). This is how you build manipulation robots fast. Capital-efficient data collection.

🕊️ TU Delft Bird Drone (Tactile Hand + Perching) — RESEARCH VIABILITY: Perching to conserve battery is novel, but this is academic research. Won't scale without funding and commercialization. Watch if any robotics VC picks this up.

🏥 KAIST Soft Robotic Hand (AI-designed, handles eggs) — GRAD SCHOOL MILESTONE: AI designing materials → AI-designed soft robotics. Cute. But materials science for robotics still doesn't have a clear moat or path to $1B+ value. Interesting, not investable (yet).

📋 YOU DECIDE: WHO WINS?

Reply to this email with your pick:

Option A: Speed + Generalization wins (Wayve beat Waymo, Skild beat traditional training). Go fast, iterate in market.

Option B: Full Stack wins (OpenAI will outspend everyone, Tesla will own robotaxis + manufacturing). Vertical integration > speed.

Option C: Legibility wins (Regulators will demand explainability, insurance will price it in). Trust > technology.

Option D: It's a tie. Different winners in different markets (consumer robotics ≠ robotaxis ≠ humanoids ≠ industrial manipulation).

Top-voted response gets featured in Tuesday's newsletter.

Hunt accordingly. ☕

See you soon,

Jose, Alberto, and the Founderscrowd editorial team