Return to site

A manufacturing field-trip to Shenzhen

By Ricky Wong

At Sinovation Ventures, we have recently put together our inaugural Shenzhen field-trip for our hardware founders, what has been a - very welcome - trend recently (see more on similar trips organized by MIT’s Media LabHAXHighway1 or HWtrek’s upcoming one). We had several of our “post seed” US-based IoT startups join the trip, including FictivFlybrixEdynDropletCompology and Securly. The week was split into group visits, customized small group discussions, and tailored 1:1 for each of the startups. Our group met with, across the board, manufacturing and packaging partners, design houses, production consultants, distributors for the Chinese market, and several local fellow hardware entrepreneurs. The trip received some coverage by the WSJ.

The various insights from the trip are covered in a series of posts below:

  1. Shanzhai, The Coming of Chinese Brands, and Crowdfunding in China.

  2. Manufacturing in China, PCB, and Assembly.

  3. Tooling, Injection Molding, and Packaging.

  4. Insights from the Heavyweights - Foxconn and BYD - and more.

  5. How to Pick a Manufacturer.

Shanzhai, The Coming of Chinese Brands, and Crowdfunding in China

One of our first stops was the famous Huaqiangbei. It’s a square mile block that sells all kinds of electronics - from end consumer products to individual components - typically at wholesale prices, that can be bought individually or by “pounds” (!).

When we were there we saw plenty of (presumably) electrical engineers bargaining with the vendors over prices. As a test, we also tried to bargain just to test the price elasticity of various products. With Ricky’s thick Hong Kong accent, we were able to get the prices to drop as much as 70% for “component” products (e.g. SD cards) when going from 1 unit to 100 units. For more “complete” products such as routers we were able to get ~15% off the retail price. One word of caution, when buying in Huaqiangbei it’s hard to tell whether you’re buying genuine parts. We were told that locals only come here to see the samples in person, for their prototypes. Some of the shops you see in Huaqiangbei are just a front-end “showroom” for a factory, somewhere far away. This is probably - actually - not the place to buy pounds of components and carry them back to your production assembly line.

broken image

We first heard of Huaqiangbei from our local hardware founders and because of “shanzhai” smartphones. “Shanzhai” is a term that is used to describe a large number of grassroot Chinese companies making knockoffs of branded products. Each company makes a slight change to the original product, sometimes resulting from customer feedback, but often driven by resource constraints.

Occasionally, the “shanzhai” ecosystem would create a series of innovative features (e.g. dual-sim cards. However, most of the products that come out of the “shanzhai” system are generally of low quality and often times comical (e.g. Hello Kitty smartphones). Having said that, because of the large number of copycats and the quick feedback loop (by being in close proximity to one another and to the back-end supply chain), the “shanzhai” ecosystem is extremely good at replicating true innovation quickly. An example is the “hoverboard” trend (or fad?) that’s all over the US at the moment, which ironically was a Chinese innovation.

broken image

Still, in the last few years, the term “shanzhai” has become a bit of a national pride. It evokes similar emotion as India’s “jugaad” or the makers of the Homebrew Computer Club. Interestingly though, at the core, “shanzhai” isn’t driven by the idealistic sense of open source sharing. “Shanzhai” comes from China’s historically weak IP protection climate and capitalistic greed. Any innovation is quickly copied and most of the value is captured by the end consumer rather by the innovator. The large number of copycat followers created a (what economists would call) perfect competition environment.

Going back to Huaqiangbei in 2015, we’re now seeing the end of the “shanzhai” smartphone movement. From our informal discussions with insiders in Shenzhen, the Chinese consumers have moved up in sophistication and brand awareness. “Shanzhai” smartphones that look like Hello Kitty or Mercedes Benz no longer attractany shelf-space in Huaqiangbei. Low cost Android devices and local Chinese smartphone brands such as Xiaomi, Meitu, or One Plus are winning market-share.

The question though is whether the “shanzhai” sub-culture will move on to other product categories or simply die. Folks we know in Shenzhen believe that “shanzhai” smartphones (at least for now) still have demand infrontier markets, such as Africa. For the Chinese domestic market, we believe we’re in a transformation phase, where Chinese consumers are now starting to appreciate branded products. By appreciation, we do not mean merely as a status symbol (as in the case of Apple’s products) because China has long been a huge market for international luxury brands, but middle-class Chinese consumers finally appreciating that “brand” is a mechanism by which companies establish trust and communicate quality with the general public. This should give rise to many brand-oriented China companies, across all categories, for years to come.

One example is CrazyBaby, one of our own portfolio companies in Shenzhen. Crazybaby recently raised $828K on Indiegogo and $536K (¥3.5M) on Taobao’s crowdfunding platform. The founder, Allen Zhang ,is a rare entrepreneur among early stage hardware startups in China, who understands both the US and China markets.

broken image

They had to do two crowdfunding campaigns, in two different time zones, and at the same time manage manufacturing. We noticed a couple of things about how the Crazybaby campaigns are done differently in each region. First, both campaigns are entirely localized. The text (obviously), videos, and many of the images are different. It’s very tempting to reuse as much media assets as possible, but the team decided to customize each campaign for the local audience. The US video is very direct. It immediately goes into sales mode and walks the user through the features of the product. The US video also used the voice over of someone with an American accent so that the customers have an easier time to understand the product. However, midway in the video, the team introduced itself in English and talked about the company’s founding story. This is quite common in many Kickstarter and Indiegogo videos, where the startup tries to rally backers behind the vision of the company and to start cultivating a community around a brand. Crazybaby’s case study is particularly interesting because it’s an example of a Chinese startup coming of age. The founders take pride of where they come from (i.e. not hiding the fact that they’re a China-based startup) and at the same time let the product quality speak for itself. Despite the political rhetorics, we find that the US consumer culture is quite open and willing to accept products from anywhere, on merit.

The Chinese video on the other hand has a slower pace. It picks a music that is much more emotional, and right from the start it is trying to get customers to buy into the broader vision of the company. The fact that the team speaks Chinese natively allows each team member to individually tell their story. The video also get into details on the product development and design decisions. It reminds us of a classic Jonathan Ive videos which are like a Hollywood theatrical trailer and where each design decision is glorified (e.g. Apple Watch andiPad Pro). This conveys to the consumer that the designers are extremely passionate about their product and they vouch on the quality. Another fact is that the Chinese campaign has many endorsements from well-known online media personalities. It is fairly common in Chinese pre-order campaigns to use such endorsements. This can include famous celebrities, well-known tech personalities, other founders, international technical experts (e.g. from the US or Japan), and of course well-known investors. Overall, the Chinese campaign has much more focus on people - both external and internal to the company.

On the flip side, prices for the two target markets are the same (~$249). The production schedule is also more or less in sync. One could make an argument in favour of shipping one region first, in order to workout the bugs, rather than shipping all regions simultaneously. Crazybaby went with the riskier strategy, but in the end they were able to ship ahead of the Christmas holiday and customers have been leaving generally positivecomments. We have generally noticed that Chinese hardware startups are able to ship relatively on-time, and often with less upfront capital. The amount of manufacturing experience and the ramp-up speed of Chinese hardware startups is also generally better. For example, what a US hardware founder would call a “prototype” is very different than what a Chinese hardware founder would call a prototype - the latter generally means that the team is ready for DFM (Design for Manufacturing).

On the other hand, the US founders and the US consumer market in general, are much more willing to try innovative (albeit unproven) products. US founders are able to quickly create a “concept” product (and in many cases just a video) and validate the market demand via crowdfunding or other means. The US consumer base is also much more entrepreneurial in the sense that it’s much more willing to buy these very early stage (or non-existent) products. If we have to choose, with the exception of highly technical products, the customer development risk is a greater risk to manage than manufacturing. Having said that, both can kill a company. Coincidentally, the week we were in Shenzhen we saw two massive crowdfunding projects hitting brick walls (Coolest and Zano).

Crazybaby is an example of a startup that can learn from both markets. It launched on Indiegogo ~10 months ahead of Taobao - presumably tapping into the early adopter market in the US. At the same time, by being in Shenzhen, it is able to mitigate its manufacturing risk.

On the topic of crowdfunding in China, our US founders also met with folks from Taobao’s crowdfunding platform and from JD. A key difference between US and Chinese crowdfunding platforms is that Chinese companies often use it as a marketing channel for quite established brands, like e.g. Hai'er, rather than for new products with genuine crowdfunding needs.

To give a sense of scale, Taobao is the C2C arm of Alibaba, which itself is a $207B company. JD is also a colossal e-commerce company at $45B. In fact, in the crowdfunding space, JD’s platform is larger thanTaobao’s platform, although Taobao is catching up rapidly because it is the overall king of e-commerce in China. By comparison, Amazon is $311B and eBay is $33B.

broken image

We had the pleasure to host Gao Zheng (head of Taobao’s crowdfunding platform) for a private discussion with our group; he shared a few general insights from Taobao’s experience. Crowdfunding in China tends to skew in first-tier cities where consumers have more disposable income. A typical persona is a male who is recently married, still young enough and with no kids, and his wife not in control of the family budget yet - i.e. plenty of disposable income to spend, and personal freedom to decide. Gao Zheng also gave some interesting examples of how Taobao is being used. For example MiniK raised $3.2M (¥21M) from 348K customers - 21X of its original goal. The product is a simple, but very cool, BLE-enabled smart-plug. The company promised shipping within 30 days of the end of the campaign, which is an aggressive goal by itself. In fact, it managed to beat expectations and shipped 2 days after the campaign ended. This is almost unheard of in the US and especially for a company that just had 21X of its original funding goal. How did this startup do it?

For one, this company already had experience building a smartplug. Their old product isn’t quite the same as the new product. The new smartplug has an industrial design that is a lot less geeky, comes with multiple colors, and has new features such as automatic prevention of overcharging of smartphones. However, it’s clear that the team has substantial experience that is directly related to this product category.

In addition, when a Chinese company enters in crowdfunding mode, it typically already has completed at least the tooling stage. MiniK saw the early traction of the crowdfunding campaign and was able to increase their order size while the campaign was still live - essentially a just-in-time manufacturing process.

In fact, in China, there are now insurance policies that companies can take out, where if a campaign is delayed, the customer will receive 30% of his / her pledge back. The customer still gets the product when the campaign ships and the 30% is a bonus compensation. Note that the best startups (the ones that are the most confident in their ability to ship on-time) would take out these policies to signal to end customers that they are confident of their ability to execute on their production plans.

This brings the question of whether crowdfunding in China is truly crowdfunding at all. Perhaps a more suitable term is simply “pre-orders”. This of course means that crowdfunding in the US is more attractive to products that are still at the conceptual stage - where the shipping date is highly unpredictable and in some cases the entire project still has substantial technical risk.

One insider told us that crowdfunding in China, unlike in the US, represents the beginning of the business - not the beginning of a dream. Chinese consumers at the moment have no interest of becoming “angel investors” of a campaign. They see themselves as consumers first and foremost.


We wonder what it would mean for startups that are trying to use crowdfunding to create an early evangelist community. In the US, you often see campaigns with a basic $1 reward as a way for consumers to buy into the early user community (e.g. the privilege to receive exclusive, supporter-only updates). In China, at first glance, you first see a similar basic reward tier. But after a closer look, these rewards are usually future coupons with substantial discounts on the final product. Part of this phenomenon can be attributed to the unique history of donation and civic culture in the US. The culture to support (“invest in”) a project just to help an entrepreneur is very much a US phenomenon.


Manufacturing in China, PCB, and Assembly

Startup founders sometimes do not understand that, just because they are the “customers”, it doesn’t automatically mean that their projects are attractive to manufacturers. In general, manufacturing is an economy-of-scale business. This is quite incompatible with the requirements of early stage startups. Startups have smaller order sizes, more discontinuous order flow, and the products are also - typically - more unusual (i.e. higher learning curve in terms of manufacturing). The future financial prospect of startups is also uncertain, which has implications on the payment terms, and on whether the manufacturer should hold materials’ inventory for the startup.

The good news is that the manufacturing sector in China hasn’t truly recovered yet since 2008. Many vendors are interested in diversifying their client base after the crash and the ensuing slowing down period. Many manufacturers recognize that, by partnering with clients that are growing fast, they themselves will likely grow fast as well. And - of course - it goes a long way if you have products that excite them, and you can also show them that you have backing from reputable investors. There are even cases where manufacturers (or their owners) would invest in the startup themselves.

Manufacturing in Shenzhen

First thing to know about manufacturing in China is that there are several regional clusters to consider, depending on the type of product that you want to build.

Just like the tech sector in the US is divided into various hubs - e.g. SV (being the main center), NYC, Boston, Austin, etc. - manufacturing in China can also be beyond Shenzhen. Nick from Brinc talked about his experience in creating a watch made of bamboo material. In his case bamboo is sourced from a specific part of China that is far from Shenzhen. He had to factor in the additional cost of transporting the bamboo to his final assembly shop, as well as the added time, to his supply-chain financing model.

broken image

Shenzhen is perhaps the most complete ecosystem in the world when it comes to manufacturing for internet enabled, electronic devices. For the purpose of this post, let’s consider what a hypothetical consumer IoT product might need in terms of manufacturing:

  • DFM / DFA / DFT

  • PCB / PCBA

  • Assembly

  • Tooling / Injection Molding

  • Packaging

Note that many Contract Manufacturers (CMs) you meet will do some combinations of the above. Others have their preferred outsourcing vendors who they’ve worked with in the past. Some would have the ambition to gradually take more tasks in-house (vertical integration) and others wish to expand by growing horizontally (taking on more clients of the same work).

For our hypothetical IoT startup, let’s assume that it has just finished a great Kickstarter campaign. However, the prototype in the Kickstarter video is simply a look-alike created from 3D printing. Our startup does have a 75% functional prototype, but it knows that it the final industrial design will look very different. It also has a BOM estimate, but it is based on the components that are in the functional prototype.

A DFX partner (DFM / DFA / DFT - Design for Manufacturability / Assembly / Test) would take the functional prototype and effectively redesign it to a large extent, based on existing components. For example, certain TI chips might be easier to use during prototyping (because of ease of sourcing and documentation), but in DFM the design partner would replace them with something that’s perhaps cheaper and easier to source at scale, or locally.

One such firm that we have visited is Axis. Axis is ran by two nice gentlemen from Hong Kong. Axis has done a lot of work for Charles Huang before (one of our venture partners and co-creator of the Guitar Hero franchise). The CEO, Jared, talked about some of the more recent projects that they’ve done with Activision. For example, in Skylanders, they found a way to enable up to four sensors being placed on the pad. This allows players to put four action figures on the platform at the same time. This technical breakthrough makes action figures more fun for the players and thus translates directly into increased merchandise revenue for Activision.

broken image

Walking into the Axis office, you’ll immediately see shelves of PCB test jigs that Jared has created for different projects. There are also machines that would automatically flip various switches (e.g. on a Guitar Hero guitar) a million times to ensure mechanical reliability. The most impressive of all is a giant machine that is the size of a huge round table to test each Skylander RFID chip. This high-speed rotating machine is able to test each RFID chip as it is coming out of the assembly line without mich human involvement. The goal is to make sure that there’s minimal interruption to an assembly line once it’s started and to remove the human element, as much as possible.

Custom machines don’t just stop at testing. Jared also showed us a machine that they created to calibrate each PS3 controller. Each controller joystick coming out of the assembly line needs to be rotated 360 degrees to calibrate for the maximum freedom of movement. It’s impossible to do this with human labor in a consistent way. Jared and his team created a machine that would automatically rotate the joysticks and record the settings into each controller’s firmware.

All the machines inside Axis are custom designed. The manufacturer would take Axis’ blueprint and create a new set for each assembly line. For each product Axis creates, they have to create 4-5 “DFX” auxiliary products to support the manufacturing and testing at the end of the assembly line. Jared also helps design the assembly process itself - how the product should be assembled, number of workers and steps in each line, and sometimes even where each worker should sit. Reducing assembly steps can have a significant impact in the overall cost of production.

PCB & Assembly

In theory it’s possible to employ an assembly shop, and then have a different vendor for PCB. In fact, we have met with vendors that are specialized in one or another in our various visits. But we have also met with three vendors that have integrated assembly and PCB capability. They are Seed, BCD, and AQS.

What else is common in all three is that they all have a Bay Area branch office, and generally have experience working with US startups. In addition, all three manufacturers are at the right size to engage with startups (i.e. not too big and willing to accept lower MOQ - Minimum Order Quantity). These are all generalist manufacturers that have experience working with a wide range of products (depending on the product type, you might want to consider engaging a specialist manufacturer, especially when rare certifications are required). At the same time, all three are very unique in their own way.

Seeed Studio has a very strong maker and hacker culture. The entire operation at Seeed is optimized for orders between 1 - 10,000 units. For example, their assembly line tends to be much shorter and each manufacturing step is generally more complex. To make it work, Seeed is at the central part of Shenzhen, which allows them to attract generally more educated workers (most assembly workers have technical or university education).

broken image

Seeed has an open module library that makers can use to build their products out of higher level abstraction. A benefit is economy-of-scale. Makers of different products that use the same underlying modules can now pull their orders together and collectively reach the MOQ requirement for various components. Finally, Seeed is known within the maker’s community as one of the top three distributors of maker products. It’s no surprise that Seeed has become a top name within the maker’s community.

BCD is run by a pair of American founders with solid Shenzhen manufacturing experience. The senior management speaks excellent English and the fact that they have a branch office in SF makes it very attractive for startups. BCD’s HQ office is in the “suburbs” of Shenzhen, but it’s accessible by a 20 min taxi ride. They recently purchased the latest model of Yamaha PCBA / Pick-and-Place machine (the best in its class). It’s always a good sign when a partner is investing in its own business.

broken image

The company is currently making a big push into wireless and IoT products. It’s reflected by the various test labs that they have created specifically for testing radio frequency interference and certifications that are maintained by the company. Overall, the entire operation at BCD feels very professional, and with startup agility.

AQS is ran by a very experienced, Korean manager. In fact, the company was originally founded in Silicon Valley back in 1991, where it still has 300 employees. What attracted us to AQS is their history of working with startups and their ability to take on wacky (but cool) projects originating in MIT. They are currently working with 10+ startups already, and it’s clear that engaging startups is a top management initiative. We recommend reading about Joi Ito, Reid Hoffman, and Marty Schmidt’s trip to AQS.

AQS has three locations in China, but the main factory is in Dongguan, which is about 1.5 hours away from central Shenzhen and where wages are ~20% lower. Depending on how labor intensive your product is, a general rule is that labor could be up to ~30% of your BOM cost.

Labor Costs

On the topic of labor costs, one useful exercise is to imagine that you’re the manufacturer. Going back to our hypothetical startup, what would it cost your manufacturer to support the startup? Let’s focus purely on assembly labor for a moment. Assume that the factory has extra capacity (i.e. not considering opportunity cost). Generally, 10-15s per task is the limit of a human’s short-term memory. For complex products, it’s common to see a manufacturing pipeline broken into 40 steps or more (i.e. 40 workers for each assembly line). Currently, the daily wage of a worker in Shenzhen is ¥120 (i.e. ~$19) or about ¥2570 monthly (~$400). But most manufacturers also provide housing and food for the workers (because most workers in factories are migrant workers who are not from the region). We can then expect the total cost of a factory worker to be ~¥3500 (~$540) monthly. This doesn’t include the cost of real-estate, machines, power, gas, management overheads, etc. In other words, if our startup wants to keep a 40 person assembly line operating for one month the factory has to spend $21,600 just on labor alone.

Of course most startups would not have the capacity to rent an assembly line for the entire month. Let’s redo our math. Suppose that we have re-engineered our DFM and we could make do with a shorter assembly line of 15 people (potentially less throughput or lower yields due to decrease in assembly complexity) and rent a line only for one day (because that would complete the entire order we got from Kickstarter) we’d then be paying ~$400 in raw labor.

In fact, there’s significant switching cost associated with setting up an assembly line and training each person to assemble a new product. Our startup should expect that the factory would charge some amount of setup fee (particularly when future orders is highly uncertain). Also note that 15s per task would mean that our throughput for each line is effectively 240 units / hour (assuming 100% yields, which is impossible). In an eight hour shift (not counting breaks and lunch) we can produce ~1920 units.

This means that we can expect $0.21 per unit in raw labor cost in the best case scenario ($400 / 1,920). One can easily imagine a scenario where this figure would increase 2X-3X due to lower throughput and lower yield. In other words, without knowing what our startup’s product is, we could already put down some boundary constraints just based on how the manufacturing ecosystem is setup.

Tooling, Injection Molding, and Packaging

In theory, tooling and injection molding can be done by two different companies. One potential benefit is that if your tools are done by one company, you could take them to a different injection molding shop. Another example is if you’re trying to make a very big part (e.g. for a car). It can be more economical to have tooling done in China and then have the injection molding done elsewhere (e.g. in Germany), instead of shipping huge pieces of plastic as they come out of the factory.

However, in most consumer electronics cases, it is recommended that you use the same company for both. Any two of the top tooling companies would design your tools slightly differently and it can be awkward for another company to re-use someone else’s tools. Furthermore, if something goes wrong with a tool during the injection process (e.g. the tool breaks) it becomes a huge blame game to figure out if the original tool maker is at fault or if the injection company is in negligence. Because of that, we recommend using the same vendor for both.

One tooling company that we visited is Sheenway. Sheenway’s tooling facility is in Dongguan and the company is run by a group of highly experienced Hong Kong managers. They have experience manufacturing for various top smart-home companies for close to 10 years (back when it was called home automation with Zwave and Zigbee). In the last few years, it has expanded into other WiFi and BLE products. One of our US portfolio companies, Swivl (a robotic device that turns your smartphone or tablet device into your personal cameraman) is currently using Sheenway.

broken image

There’re a number of considerations when it comes to tooling. For most startups, you would care about size, precision, and strength. They all have cost implications. Obviously, the larger the part, the more expensive it is. But sometimes for smaller parts, in order to increase injection throughput, you can put multiple parts on the same tool. For example, you can put four smartphone plastic parts on one tool, so that each injection will create four pieces. The assembly worker would then break each plastic part out, after it is injected.

broken image

Precision of the tool generally has to do with the quality of the machine that makes the tool (see this video on the machines involved in making a tool). A top end German or Japanese tool-making machine can easily cost $100K-$200K and factories would let these machines run 24/7 (i.e. with a night shift to keep an eye out for them) to amortize the capital cost. The strength of the tool determines how many times you can use it to create parts. You can generally expect that a decent tool can inject at least 500K units before breaking.

For a small part such as a smartphone back piece, you can expect to spend ~$10K on tooling and ~2-3 months of lead time. This includes the time for the tooling shop to design (DFM) the tool and have it ready for the first injection molding production run. Of course a lot of these numbers are just a rule of thumb. We’ve even heard of a well known consumer electronics company spending close to $1M on tooling (it’s a very complex robotic product with many parts).

Needless to say, because of the expensive upfront cost, it’s highly advisable that you do not change your tools often. The tool determines the size of the product, which has implications on the size of your PCB and your packaging. To make it worse, tooling has a long lead time. This combination of challenges means that you have to somehow start your tooling process early and manage the rest of the DFM constraints in parallel.

A lot more can be discussed on tooling and injection (e.g. injecting different kinds of materials such as rubber and metal, double injection, adding texture, stopping deformity, tonnage for different sizes, etc.). Many of these issues depend on the nature of your product.

It’s important to think about assembly and how easy it is to put the parts together. One good resource isFictiv’s hardware teardown series (e.g. for BB8, Nintendo, or Muse). The ideal case is that your industrial designer from day one has some sense of tooling and the various trade-offs in the process. At the very least, do not think of tooling and injection molding isolated from the rest of the manufacturing process.


Packaging is quite important (and usually underestimated by founders) in the US market because the offline retail channel is still a big sales channel for hardware startups. In a way, packaging is its own product. Packaging has its own process in design, prototyping, manufacturing, and testing (e.g. drop testing).

broken image

For packaging, we met with the team at Yuto, which is trusted by many of the world’s leading brands and typically works with high volume customers. We also met with Adam Melton, CEO at Green Packaging. One interesting thing about packaging is that it goes through fashion trends. For example, Green Packaging started when using recycled materials was just in its beginning. Green Packaging also offer a design service that can keep your product’s unboxing experience modern, fresh, and within budget. They can also help with retail fittings, or exhibition stands.

Insights from the Heavyweights - Foxconn and BYD - and more

Shenzhen is home to a number of well-known international companies. Being the capital of consumer electronics’ manufacturing, we met with a couple of the heavyweights of the industry - Foxconn and BYD, and more.

Generally, we don’t recommend startups to engage with a large, top-tier manufacturers because of scale mismatch. But some of these companies have, in recent years, made a solid effort to engage with startups.

PCH is perhaps the furthest ahead. Their Highway1 startup accelerator is generally recognized as one of the top hardware focused programs in the world (together with HAXBolt and Lemnos Labs). One of our US portfolio companies (Drop) has been using PCH as their manufacturer and has been available at Apple retail (offline and online) for more than a year now.

Foxconn is the manufacturing powerhouse ($40B market cap) in the industry. Within Foxconn, there’re many divisions that are run very independently of each other, and each has its own P&L responsibility. The group we visited is iDSBG and it is the manufacturer of one of our China based portfolio startups - AiNemo. AiNemo is building a robotic platform to help connect the modern, distributed family. It’s great to see that a manufacturing heavyweight can “dance” with the agility of a startup.

broken image

Entering Foxconn’s campus is quite an experience. Foxconn has multiple “mega factories” in China. Each feels like a small city - with their own restaurants, hospitals, schools, security force, and even marriage match-making service. Foxconn takes security very seriously. There are many sites where we had to have the cameras on our phones duct taped. Every building has full airport level body scanners and glass doors. It certainly instills confidence into a customer that Foxconn takes supply-chain integrity seriously. It’s no wonder that Apple trusts Foxconn to make its iPhones.

BYD is a Chinese entrepreneurial success story. The company was started in 1995 and had its IPO seven years later. BYD has three main divisions - batteries, cars, and consumer electronics. It is generally more well known as a maker of batteries and electric cars. BYD has perhaps become well known in the West after Warren Buffett made a historical investment in 2008. The consumer electronics division recently began an effort to engage with startups.

broken image

In some ways, BYD is similar to Foxconn. Its factory campus also feels like a small city. The company is making a serious effort to attract startups and innovation. In fact, for both Foxconn and BYD we have promised that we won’t disclose the content of our private visits. They both have state-of-the-art manufacturing facilities.

We also met with Shipeng Li, CTO of / The company is one of the largest online marketplaces to connect suppliers of IC circuits and electronic components with SMEs that are looking for parts. Shipeng, who was a founding member of Microsoft Research Asia, talked about their plans on creating a unified IoT platform. The company already has an office in SF and is giving startups power to source components just like big companies do.

Ingdan has also created a bit of a retail showcase on its ground floor, in the central part of Shenzhen. It is a good reminder why we’re all doing this - because of the cool products that are coming out of startups.

broken image

We also visited DJI’s HQ. DJI needs no introduction. It owns 70% market-share of the global consumer and professional drone market and it’s beating high-end players from the US and low-end copycats from China at the same time. We had a private discussion with Benz Huang who heads a lot of their developer efforts. While our conversation at DJI was private, we left Shenzhen feeling that the company culture is much more like Apple’s rather than Google’s or Facebook’s. Frank Wang (the founder and CEO) is a lot like Steve Jobs, in the sense that the company is driven by a visionary founder.


How to Pick a Manufacturer

How to Pick a Manufacturer

A final note. In general, it’s quite difficult to change manufacturing partners. We’ve all heard many manufacturing horror stories during our short Shenzhen trip, and some of our own portfolio startups have experienced switching partners before, and it was never easy. Therefore, it’s best - as much as possible - to pick a good manufacturing partner from the start, rather than “iterate” into a good partner.

Here is a list of general advice we’ve gotten on how to find a good partner:

  • Someone who is happy with your MOQ (Minimum Order Quantity). Manufacturing is an economy of scale business. Match your startup’s scale with the manufacturer's scale and grow together.

  • Someone who is flexible and willing to change their assembly line process for you. Let you have input into how assembly should be setup.

  • Figure out how they manage their First-In First-Out (FIFO) inventory. They should be able to show you clearly.

  • They have a system to track their orders.

  • Meet the owners of these companies from day one and establish personal rapport. The Chinese culture is quite interpersonal and signed contracts are simply considered as a starting position in a negotiation.

  • See factories' previous products and how everything works. Meet their lead mechanical and electrical engineer when you visit.

  • Check out how factories conduct QC (quality check) and packaging at every step. Hire boots on the ground to inspect that process for you (if you are not based in Shenzhen yourself). Startup founders should make a personal trip to their CMs at least once every three months, and their local agents should remain in constant touch with the CMs.

  • For plastic injection molding companies, check how they store their tooling. It shows how good the management is, which leads to final product quality.

  • Someone who you can communicate effortlessly with.