Daily Pulse #54 | Aug 3 Monday Edition | Unitree's STAR Market IPO subscription opens Aug 10 — China's first humanoid robot stock; BYD to unveil its first humanoid robot this month; the $30/hour cleaning robot is actually teleoperated

🎯 Monday Briefing (Aug 3)

The new week opens on a gap between capital and reality: Unitree Robotics’ STAR Market IPO registration was approved, with subscription opening August 10 and RMB 4.2 billion to be raised — China’s “first humanoid robot stock” is in its final countdown. BYD will unveil its first humanoid robot this month, deploying it in Zhengzhou, as the automaker race into robotics intensifies. On the other side, San Francisco’s viral $30/hour home-cleaning robot has been confirmed as “human teleoperation + AI assistance” — the “robots doing the work” filter is being peeled back, layer by layer.

• 🏦 Unitree IPO countdown — subscription opens Aug 10, raising RMB 4.2B, backed by Meituan and Alibaba; A-share “first humanoid robot stock” • 🚗 BYD humanoid this month — first model unveiled in August, Zhengzhou deployment, commercial service scenarios first • 🤖 The $30/hr cleaning robot is teleoperated — San Francisco’s viral home-cleaning service is “pure human + intelligence” • 📦 DoorDash pays people to load its delivery robots — automation’s last mile still needs humans • 📊 Barclays: Automation 3.0 era begins — humanoid robot market could reach $200B in a decade

1. 🤖 San Francisco’s $30/Hour Cleaning Robot Goes Viral: Behind “Robots Doing the Work” Is Teleoperation

In one line: A San Francisco company’s home-cleaning service has gone viral — humanoid robots clean homes at about $30/hour, and they even “wash their hands before starting work”; the catch: everything is still controlled remotely by human operators, with AI only assisting.

💡 Why it matters: This is a landmark “de-mystification” case for embodied AI. The “robot butler” of marketing narratives is actually a hybrid system of “remote human operators + robot body + AI assistance” — cost structure, capacity limits, and service pricing are all decided by humans sitting at screens. The Chinese coverage got it in one phrase: “pure human + intelligence” (纯·人工·智能). When customers pay a premium for “a robot coming to your home,” what they’re actually buying is a telepresence for remote labor.

Key facts:

Pricing: ~$30/hour (200 RMB), above typical housekeeping, monetizing the “robot experience” premium
Operating model: Human operators teleoperate the robot to perform cleaning motions; AI handles auxiliary tasks like path planning
Industry signal: Multiple companies are taking the same route — validate demand with humanoid robots + teleoperation first, then raise the autonomy ratio
Why it matters for the industry: Teleoperation is the “intermediate form” of embodied AI at scale — and the best channel for collecting real-world operation data

📌 Takeaway: The “robots replace humans” narrative needs a discount: most humanoid robot services today are “remote humans in a robot skin.” But for the industry, this is exactly when the data flywheel starts spinning — every teleoperated session feeds the training data that future autonomy will run on.

🔗 36Kr (Chinese) | Sina Tech (Chinese) | ABC News

In one line: Unitree Robotics’ STAR Market IPO registration has been approved, with online subscription set for August 10 — raising RMB 4.2 billion, with Meituan and Alibaba among its shareholders, putting the A-share “first humanoid robot stock” (and “first embodied AI stock”) into its final countdown.

💡 Why it matters: From application acceptance on March 20 to listing review on June 1 in just 66 days, and now subscription on August 10, Unitree has moved through the STAR Market process at exceptional speed. With 2025 revenue of RMB 1.70 billion, RMB 591 million in net profit (excl. non-recurring items), and roughly 60% gross margin, it is one of the few general-purpose robot companies worldwide that is profitable at scale. Its IPO pricing will set the valuation anchor for the entire humanoid robot sector.

Key points:

Raise: RMB 4.2 billion, earmarked for capacity expansion and R&D
Shareholders: Meituan, Alibaba and other internet giants — clear strategic coordination intent
Fundamentals: Over 5,500 humanoid robots shipped in 2025, the most worldwide; humanoid revenue share has overtaken quadruped
Market impact: The post-listing valuation will define the pricing reference for embodied AI startups in private markets

📌 Takeaway: The subscription and listing of the “first humanoid robot stock” is one of the most important capital events of 2026 for robotics. It moves the sector from “narrative phase” to “pricing phase” — and the post-listing share price will directly test the market’s real expectations for embodied AI commercialization speed.

🔗 Sina Finance (Chinese) | iFeng (Chinese) | Wallstreetcn (Chinese)

3. 🚗 BYD to Unveil First Humanoid Robot This Month: The Automaker Race into Robotics Heats Up

In one line: BYD plans to release its first humanoid robot this month, deploying it in Zhengzhou, with commercial service scenarios likely landing first — another major automaker entering humanoid robotics after Tesla and Xpeng.

💡 Why it matters: The automaker logic for building robots is self-consistent: reuse mature supply chains, factory scenarios, and production-line data. BYD’s differentiation is “commercial service scenarios first” — not showing off tech first, but finding revenue-generating deployment scenarios first. The Zhengzhou deployment signals a path aligned with local industrial policy, consistent with BYD’s classic “production-led R&D” style.

Key points:

Timeline: First product unveiled in August — among the faster commercialization timelines in the automaker camp
First scenarios: Commercial services (reception, guidance, delivery) rather than factory work
Industry context: Tesla Optimus, Xpeng Iron, NIO, Xiaomi are all in — “automakers building robots” is now table stakes
Supply chain edge: Motors, reducers, batteries, and manufacturing processes overlap heavily with automotive supply chains

📌 Takeaway: The influx of automakers means the “mass-production capability” contest in humanoid robotics starts early. Compared to startups, automakers bring a killer combination of supply chain management, cost control, and scale manufacturing — in the second half of 2026, the competition shifts from “who can build one” to “who can build it cheap.”

🔗 CLS (Chinese) | Guancha (Chinese) | National Business Daily (Chinese)

4. 📦 DoorDash Pays People to Load Its Delivery Robots: Automation’s Last Mile Still Needs Humans

In one line: Business Insider reports that DoorDash is paying human workers to load its delivery robots — the robots do the curb-to-door leg, humans do the loading, a division of labor that exposes the real weakness of automation at the “last mile.”

💡 Why it matters: Delivery robots solved “the road” but not “the loading”: pick-and-place of arbitrary shapes and orientations is the hardest part of general manipulation. DoorDash’s choice is pragmatic — spend the minimum on human labor for the hardest part rather than wait for general-purpose robot manipulation to mature. Commercially, this confirms: perception and locomotion have crossed into early commercial use, but dexterous manipulation remains embodied AI’s biggest bottleneck.

Key facts:

Division of labor: Human employees load robots at stores/warehouses; robots complete outdoor delivery
Business logic: Paid human labor has the best ROI today compared to waiting for general manipulation tech
Industry-wide pattern: Many delivery-robot companies use similar “human-robot relay” models

📌 Takeaway: Together with the cleaning robot story, this points to the same truth from two markets: commercial robot deployments in 2026 are generally “robot body + human backup.” Dexterous manipulation is the hard problem that must be solved next — and the dividing line for the next round of valuation divergence.

🔗 Business Insider

5. 📊 Barclays: The Automation 3.0 Era Begins, Humanoid Robot Market Could Reach $200B in a Decade

In one line: Barclays’ latest research introduces an “Automation 3.0” framework — a new wave of automation led by embodied intelligence is beginning, with the humanoid robot market projected to reach roughly $200 billion in 10 years.

💡 Why it matters: This is the first time a major investment bank has used “Automation 3.0” (distinct from 1.0’s fixed industrial arms and 2.0’s AGV/AMR mobile robots) to define the embodied intelligence era — a sign that sell-side research is pricing general-purpose robotics as an independent industrial cycle. The $200 billion target implies roughly 40% CAGR — more measured than the market’s “100x expansion” hype, but directionally confirming the sector’s long runway.

Core logic:

Stage framework: Automation 1.0 (fixed arms) → 2.0 (mobile robots) → 3.0 (embodied/general-purpose intelligence)
Market forecast: Humanoid robots could reach ~$200B within a decade
Drivers: Labor shortages, AI foundation-model capabilities spilling into the physical world, rapidly falling hardware costs
Risks: Dexterous manipulation, data acquisition, and safety certification remain the three bottlenecks to commercialization

📌 Takeaway: Sell-side research moving from “theme hype” to “industrial cycle” framing is a sign the sector is maturing. But the $200 billion prize depends on “Automation 3.0 actually working” — and as the first two stories show, the most profitable robot businesses today are “human-robot hybrid,” not “fully autonomous.”

🔗 CLS · Overseas Research (Chinese)

🔍 Monday Market Watch

Today’s five stories form a panorama of the humanoid robot industry:

Capital: Unitree’s subscription opens Aug 10 — the “first humanoid robot stock” lands imminently, and the sector enters its secondary-market pricing phase; the valuation anchor is about to be set.

Supply: BYD unveils its first humanoid robot this month; the automaker camp is all in — mass-production capability and cost control become the new competitive dimension.

Demand: San Francisco’s $30/hour cleaning robot went viral — consumers will pay a premium for “a robot at the door,” but the truth is teleoperation. Whether the “robot experience” premium holds depends on how fast the autonomy ratio climbs.

Bottleneck: DoorDash pays humans to load robots; cleaning robots run on remote operators — dexterous manipulation is the shared weakness of embodied AI and the dividing line for the next round of differentiation.

Research: Barclays frames “Automation 3.0” with a $200B long-term horizon — sell-side is pricing embodied intelligence as an industrial cycle, not a theme.

One-line summary: Capital is front-running, automakers are flooding in, scenarios are being validated, and bottlenecks are being exposed — humanoid robotics is moving from “storytelling” to “counting the pennies.”

Sources: 36Kr/QbitAI, Sina Finance, iFeng, Wallstreetcn, CLS, Guancha, National Business Daily, Business Insider, ABC News