Related Articles
Stay Up To Date
Something has gone wrong, check that all fields have been filled in correctly. If you have adblock, disable it.
The form was sent successfully
After recent trips to China, Sands Capital Sr. Research Analyst Massimo Marolo, CFA, finds China’s industrial backbone combined with proactive policies around market entry, R&D incentives and antitrust oversight, create fertile ground for automation to redefine productivity and competitiveness. In this series, On the Road, we share the findings uncovered by our on-the-ground research.
Key Points
- China’s current landscape reflects a tension between an economy still finding its footing in a consumer-driven era and a manufacturing sector racing toward an AI-powered future.
- While near-term pressures on domestic consumption may temper broader growth, the durability of China’s industrial backbone and its deliberate fostering of innovation suggest that the robotics renaissance is more than a cyclical fad.
- Over the next decade, investors who target specialized automation equipment makers, precision component suppliers and emerging technology champions stand to benefit disproportionately.
I’ve been going to China once or twice a year since 2010, and there’s always something interesting going on. This time, I was struck by the disconnect between an economy still struggling in its transition to consumer-driven demand—and the dizzying array of choices created by technologies sitting at the intersection of AI and robotics.
One part of this has to do with China’s existing manufacturing ecosystem.
China makes about one-third of the world’s goods. Manufacturing, at about 15 percent of global GDP, remains the heavyweight driver of global growth. By comparison, information and communication technology (including software) accounts for just 6 percent of global GDP. It is logical to think that AI’s most powerful manifestation will come from its application to the manufacturing of things via advanced robotics. With a 50 percent share of all new industrial robot installations (400,000 units per year versus the United States at 45,000), and a 40 percent share of cumulative robot installations (versus #2 Japan at 10 percent), there’s a credible argument that China’s legacy advantage in this realm will be long-lasting.
The density of China’s components ecosystem is also relevant. Taking Tesla as an example, the local content ratio at its Shanghai’s gigafactory is 95 percent versus 60 percent and 55 percent at its Berlin and Fremont plants. For the iPhone, Shenzhen hosts 286 supplier sites run by 157 firms within a 100-kilometer radius. Alibaba’s domestic B2B platform 1688.com lists 1.5 million manufacturers and suppliers. Made-in-China.com, an export facing directory popular with European Union and U.S. purchasers of machine parts, lists 6 million registered supplier accounts, representing 70 million product listings across 27 industries— more than the combined manufacturing sector of many G20 economies. For robotics, it is estimated that per-unit costs are 20 percent to 40 percent lower than in the United States and EU.
The result is a lot of competition and consumer choice. It is noteworthy that on a purchasing power parity basis, which considers the relative cost of goods, China’s economy is currently 35 percent larger than that of the United States ($40 trillion versus $30 trillion). Chinese consumers can choose from among 50 brands of domestic electric vehicle (EV) manufacturers (versus eight in the United States), 16 brands of TVs, 32 brands of smartphones, 20+ sneaker brands, and so forth and so on. It is a similar picture at the enterprise level, with 27 industrial-robot manufacturers, 105 makers of lithium-ion batteries, 400+ makers of photovoltaics, etc.
What is also happening is that novel “Made-In-China” technologies, namely EVs and lithium-ion batteries, are beginning to set industry standards globally. I always wondered why, despite its ability to launch space rockets and build aircraft carriers, China never introduced a credible commercial jetliner. The answer is not as much about technology as about being a first mover in setting global standards. Passenger jets are truly global. They land in, and take off from, every country. This requires international regulations (International Civil Aviation Organization—which sets standards implemented by FAA) and a vast network of certified maintenance parts, both of which represent a massive barrier to entry for a newcomer wanting to muscle into Boeing and Airbus’ duopoly. EVs and lithium batteries might not require the same level of regulatory standards as jetliners, but this might be the first time that a big global industry takes shape along standards set by Chinese entities. The economic and geopolitical ramifications are worth thinking about.
A second part of this has to do with the slower, more uneven path to industry consolidation in China.
Fierce competition is rooted in Chinese culture. Some of this is by design and perhaps an integral part of the Chinese Communist Party’s (CCP) social contract with its citizens in exchange for maintaining a one-party system. We have to consider whether the consumer “abundance” can remain part of the party’s tool kit in an era in which AI begins to replace many jobs, especially in manufacturing, in which 120 million are employed?
Starting a business in China has become incredibly easy. A single online form takes care of all business licenses and permits. Paid-in capital requirements have been cut from RMB30,000 to RMB1. Central budget reimburses local governments for rent rebates, cloud credits, and patent-filing fees. Firms in strategic emerging industries get a 200 percent deduction for R&D spending. VAT for small tech firms was cut to 3 percent. So-called state guidance funds raised over $1 trillion by 2024 to invest in over 2,000 VC funds, with yearly disbursements thus far in the order of $50 billion. Broader new business formation is 23 registrations per 1,000 people, approximately 40 percent higher than in the United States. There is a long list of policies aimed at spurring entrepreneurship and competition.
On the flip side, the government actively “manages” consolidation. Some industries have reached similar concentration levels as in the United States (food delivery /ride hailing / B2C ecommerce all display concentration ratios (CR3) north of 80 percent) but operate under far greater scrutiny. China’s anti-monopoly law sets “market dominance” at 50 percent market share (or 66 percent if split between two companies), triggering regulators’ authority to undertake corrective measures. These ranges from algorithm tweaks and fee cuts to cease-and-desist orders to full asset break-ups. Recent examples include China’s Ministry of Transport ordering all ride hailing apps to publish real-time take rates and lower what it deemed excessive commission fees. Similar measures took place in food delivery with selective caps on restaurant take rates. Since 2021, the State Administration for Market Regulation has fined platforms over $3 billion for commission-related abuses, approximately four times more than equivalent government fines in the United States.
Throughout China’s economy one can find evidence of industry consolidation being tougher to achieve. Not to say that it won’t happen overtime. It has, in fact, already been happening for a few decades, but all the clues suggest that it is by design more actively managed, quite possibly for the benefit of the consumer as part of the CCP’s implicit social contract.
Industry | China Concentration | U.S. Concentration | Key Policy / Market Notes | ||||
|---|---|---|---|---|---|---|---|
Cement | CR5 = 30% | CR5 = 65% | Plant-capacity caps limit consolidation | ||||
Steel (CR3) | CR3 = 30% | CR3 = 73% | Plant-capacity caps limit consolidation | ||||
Beer | CR5 = 45% | CR5 = 85% | No formal caps; premiumization opens room for independents | ||||
Confectionary | CR5 = 45% | CR5 = 70% | Fragmented due to diverse local brands | ||||
Food Retail | CR5 = 25% | CR5 = 60% | Highly fragmented with many regional and mom & pop chains | ||||
Hotels | CR5 = 40% | CR5 = 65% | Multiple domestic chains encouraged, foreign brands face limits | ||||
Movie Theatres | CR5 = 30% | CR5 = 70% | Nb of screen quota system keep chain share in check | ||||
Source: Sands Capital research as of June 2025. For illustrative purposes only. CR = concentration ratio | |||||||
Automation and Humanoids
More details below, but the punchline is that harmonic drive systems (HDS)—the precision gear systems that allow robots to move smoothly and accurately—remain my favorite humanoid robotics hardware plays, even though I still question whether hardware is the most rewarding part of the value chain in this theme.
Japanese companies seem to lead the competition in this space, but per above points, there will always be a long tail risk that Chinese competition will gradually catch-up. This has not yet happened because the global total addressable market (TAM) for harmonic strain wave reducers is tiny at $1 billion. However, we expect it to materially inflect up and our estimates call for a 20 percent compound annual growth rate (CAGR) out to 2030 in the TAM for all light payload harmonic strain wave reducers, with the much smaller portion attributable to humanoids sustaining a 200 percent CAGR. Still, humanoid-related TAM for harmonic reducers won’t amount to more than $250 million by 2030.
Beyond HDS, the consensus among industry participants is that there are a few hurdles to overcome before general purpose humanoids become more widely accessible. These are: 1. turning camera pixels into action-ready information fast enough, 2. collecting the billions of “real-world experience frames” needed to make those perceptions reliable, and 3. blending slow, language-level reasoning with sub-millisecond reflexes that keep a humanoid from falling (i.e., slow brain versus fast reflex). Smaller hurdles include the cost of precision actuators (hence the attractiveness of HDS) and reliable hand manipulation.
On 2., per the slide below from UC Berkeley’s Chief Robotic Engineer Ken Goldberg’s presentation, we’ve so far accumulated only 10,000 hours of real-world robot interaction data versus some 100,000 years of human-generated data that Alibaba’s Qwen LLM can scrape from the open internet. Bridging that gap is no small task. NVIDIA’s Jensen Huang recently spoke about using AI to synthetically amplify and thus accelerate real world human demonstrations currently used for training. Yet, real-world training, while slow and costly due to hardware fatigue, is essential for unscripted human social interaction and environmental factors that can affect object manipulation (changing ripeness of a fruit or how micro-fissures propagate on an eggshell under irregular finger grasp).
Solving this challenge is attracting lots of newcomers with some 25 Chinese humanoid companies versus six in the United States. One company, Engine AI, claims it can compress the timeline to general purpose down to a few years. The company was recently recognized by NVIDIA and arguably has the most credible humanoid hardware on the market. Its full-size SE01 stands at 170 cm, weighs 55 kg and packs 32 joints (i.e. harmonic strain wave reducers). It walks at 2 m/s using an end-to-end neural gait controller running on a Jetson Orin + Intel CPU stack, backed by six HD cameras, a RealSense depth sensor, and a 360° LiDAR head. Its little brother, the PM01, is 1.38m /40 kg and was the first ever all electric humanoid to land a front-flip. It retails in China for $12,000 to $14,000, far below Western prototypes. Interestingly, they develop their own in-house harmonic reducers, but it doesn’t seem like they’re buying from HDS.
China’s national data backbone does potentially give local competitors an advantage in training. Beijing’s 2023 “Humanoid-Robot Innovation Guideline” calls for a nation-wide test-data platform, allowing firms, such as Engine AI to pretrain vision models on footage from the world’s largest surveillance grid (600 million+ CCTV cameras). In Shenzhen, PM01 humanoids patrol pedestrian zones beside human officers, greeting tourists via DeepSeek LLM, and flag lost children or persons of interest with on-board face detection. Engine AI has raised roughly U.S. $42 million to accomplish all this, a tiny amount next to Figure AI and Tesla.
All that said, humanoids are moving out of the lab but remain in the “preindustrial” phase. Chinese experts project only 25,000 domestic units in 2026, climbing to 220,000 by 2030, still orders of magnitude below industrial-robot volumes. The next few years (2026/2027) will likely be about narrow scope / controlled environments, such as concierge greeters. Moving toward 2030, we’ll start seeing an inflection in “task cluster” humanoids within factory settings (sorting, low precision assembly) as component prices fall and safety standards mature. Post-2030, we could see the emergence of large-model controllers enabling true general-purpose manipulation, but still bounded by economic ROI versus specialized automation as many tasks will always be best handled by purpose-built robots (i.e., no need for a humanoid to sweep the floor, or no need for humanoids to move an entire city block.)1
Narrow-focus robots from concierge help desks to cleaning bots are where I believe the action will be for the foreseeable future. However, it will be hard to make a high conviction bet on hardware. Most companies here are still private. We will be watching public cobot companies for investment opportunities.
Racing Toward an AI-Powered Future
China’s current landscape reflects a profound tension between an economy still finding its footing in a consumer-driven era and a manufacturing sector racing toward an AI-powered future. The sheer scale of China’s production capacity—responsible for one-third of global output and half of new industrial-robot installations—combined with proactive policies around market entry, R&D incentives, and antitrust oversight, creates fertile ground for automation to redefine productivity and competitiveness. While near-term pressures on domestic consumption may temper broader growth, the durability of China’s industrial backbone and its deliberate fostering of innovation suggest that the robotics renaissance is more than a cyclical fad. Over the next decade, investors who target specialized automation equipment makers, precision component suppliers, and emerging technology champions stand to benefit disproportionately. A measured, long-term approach will be key to capturing China’s unfolding automation story.
1 https://www.scmp.com/video/technology/3313381/hundreds-robots-move-entire-city-block-china)
Disclosures:
The views expressed are the opinion of Sands Capital and are not intended as a forecast, a guarantee of future results, investment recommendations, or an offer to buy or sell any securities. The views expressed were current as of the date indicated and are subject to change.
This material may contain forward-looking statements, which are subject to uncertainty and contingencies outside of Sands Capital’s control. Readers should not place undue reliance upon these forward-looking statements. There is no guarantee that Sands Capital will meet its stated goals. Past performance is not indicative of future results.
All investments are subject to market risk, including the possible loss of principal.
Differences in account size, timing of transactions, and market conditions prevailing at the time of investment may lead to different results, and clients may lose money. A company’s fundamentals or earnings growth is no guarantee that its share price will increase. Forward earnings projections are not predictors of stock price or investment performance, and do not represent past performance. Characteristics, sector (and regional, country, and industry, where applicable) exposure, and holdings information are subject to change and should not be considered as recommendations.
The specific securities identified and described do not represent all of the securities purchased, sold, or recommended for advisory clients. There is no assurance that any securities discussed will remain in the portfolio or that securities sold have not been repurchased. You should not assume that any investment is or will be profitable.
As of June 24, 2025, NVIDIA was held across Sands Capital strategies and was mentioned in reference to its founder’s comments about using AI to synthetically amplify and thus accelerate real world human demonstrations currently used for training humanoids. Other companies mentioned are not held in Sands Capital strategies and were cited to illustrate the views expressed in the commentary.