Happy new year! I kicked off the new year traveling in Shanghai, Jingdezhen, Shenzhen, and Zhongshan, each city bringing unique flavors of cool things happening in China. Shanghai and Shenzhen to learn about the tech ecosystem, Jingdezhen for a pottery retreat, and Zhongshan for seafood and chill vibes in my home city.
What’s new for Soft Patterns in 2026? You will start seeing short essay series, where I go deep on one idea across 2–3 posts. We ended 2025 with the idea of an exploration loop, first sparked by Sapiens. This post is part two of the series: a look at how compounding loops show up in Shenzhen and why the Chinese tech ecosystem operates so differently from the Silicon Valley model.
From factories to hardware R&D labs
In December, we explored the thesis that players with an exploration loop between capital, science, and an integrated stack (energy, data, and distribution) get more advantage because they can fund long cycles of exploration and translate them into habit-forming products at scale.
How does it apply to Shenzhen? The city is known for its physical AI industries (robotics, drones, smart EVs), with a prototype-to-production loop that is typically weeks faster than Silicon Valley’s.
This speed comes from two feedback loops: prototyping speed (0 to 1) and industrialization speed (1 to 100K). While a Silicon Valley engineer might wait 4-6 weeks for custom parts, a Shenzhen engineer can walk to Huaqiangbei and return with a rare sensor or a custom battery in an afternoon. This is possible because of the easy access to “unlisted” components and subassemblies that aren’t even in a catalog yet.
The other loop is designing for the manufacture floor. In Silicon Valley, an engineer may design a beautiful CAD that is unbuildable. In Shenzhen, the engineer is literally told by the factory boss on Tuesday why their screw placement will slow down the line on Friday. That iteration in small increments is a key compounding loop; Silicon Valley engineers usually "pay that tax" at the very end of the cycle. This co-location of R&D and manufacturing within a 1-hour radius turns the city into a living lab.
Hardware legos lower the cost of being curious
In most markets, commoditization (falling price, standardized parts) is seen as a threat to margins. Every hardware innovation has a "clock": the moment you ship, the clock starts ticking until competitors figure out the "how" and margins drop from 100% to 5% (from a one-of-a-kind innovation to everyone can buy it cheaply on Amazon).
In Shenzhen, commoditization is a “feature.” Because the city built the world’s smartphones, it created a thriving marketplace of “standard parts.” Specialized tech components (like a brushless motor or camera module) are now so cheap that the cost of being curious with them is near zero.
Beyond the low cost per learning cycle (how much did it cost to realize the sensor was wrong?), this also makes “kit-bashing” possible: grabbing modular components and building a new prototype in weeks, like legos. Engineers don’t need to reinvent a battery connector to innovate on the drone's flight logic, allowing them to plug-and-play and focus entirely on the end-to-end experience.
This combining with a large domestic market in China with dense engineering talent, Shenzhen operates at the high-velocity spectrum of the exploration loop and it comes with an innovation model with a “Shanzhai” and hyper-competitive culture. This is very different from the Silicon Valley model, which is typically high-risk, high-reward, and highly academic, driven by multi-year VC-funded bets on world-changing software.
Who benefits most?
Shenzhen is great at scaling and mass production once the building blocks exist (e.g., making the most affordable, AI-powered drone) (1-100K). Silicon Valley is great at creating new categories from scratch (e.g., creates the first LLM) (0-1). As AI interacts with the physical world, a faster exploration loop for rapid real-world testing becomes critical beyond just making things cheaper.
What kinds of business benefit the most from a Shenzhen loop? Consumer electronics is the iconic category. Smartphones, drones, wearables, robotics, foldable tech are lifestyle accessories rather than life-critical tools (e.g., medical devices or automotive), and they thrive on a “ship-and-patch” culture. Teams use Shenzhen’s supply chain density to release slightly raw, buggy hardware to early adopters, using the market as a real-world focus group to inform the next iteration in weeks.
From a market lens, three things came out of it: (1) commoditization of hardware components lower innovation barrier for smaller teams without a $100M R&D budget (e.g. companies like DJI and insta360 exist because smartphone components like GPS, cameras were commoditized first). (2) create complementary/accessory markets bc it leads to standard interface (bluetooth, USB-C), which creates a massive, predictable market for accesories (e.g. commoditization of smartphone created the multi-billion dollar case and charger industry), (3) dominant player can force commoditization on a niche to outcompete smaller players who cannot survive on thin margins, leaving the bigger player as the last one standing with the most efficient scale.
When does the loop fail?
This loop breaks when a company’s value comes from something that can’t buy off a shelf:
Proprietary physical data: A drone company shouldn't just be iterations of motors, but the iterations of navigation data collected by those motors. The advantage is the edge-cases the AI has learned because the hardware was cheap enough to deploy at scale.
Taste and status: The “cool” factor. People buy a Leica because of how it feels and what it says about them, not because it has the cheapest sensors.
Deep tech: Scientific breakthroughs. You can't kit-bash a quantum computer; it takes years of lab work that speed alone can't solve. However, companies like Unitree Robotics blur this line, where they took "impossible" humanoid robotics and, through the Shenzhen loop, made one for the price of a used car.
Platform ecosystem: The 'sticky' factor. Apple doesn't have to be first because once you have the phone, the watch, and the laptop, leaving is too much of a headache.
What I’d watch
Cost of learning cycle: I’ll bet on teams that launch “rough” versions and fix them in days based on what customers actually say or that it’s positioned as a player that differentiates with things that can’t be bought off a shelf (e.g., taste-driven brands, deep tech, platform ecosystem).
Avoiding commodity trap: For Shenzhen teams working on hardware AI products, I’ll watch as they scale, do they slow their exploration velocity? Assess whether they start developing advantage beyond iteration velocity, such as identifying real-world edge case through propriatory physical data, building their software suites of tools that retain users, etc.
Design for manufacture: For Silicon Valley teams working on hardware AI, I’ll watch their iteration velocity over six months and whether there are collaboration frictions that slow their exploration loop since US designers may be miles away from the assembly line in Shenzhen or Asia.
Global data compliance: I’d watch how early a team integrates global data standards and use parts that is compatible with Europe/US. A high-velocity loop that ignores compliance will hit a wall when trying to scale beyond the domestic market.
In the next post, we will deep dive on a Shenzhen startup as an iconic example that used this high-velocity loop to outcompete industry leaders in the consumer electronics market. Stay tuned!