4 ms·

"It would be better just to build a Shenzen in the US" Probably not. Imagine if you went to a flea market, but instead of comic books, hot dogs and used sofas

"It would be better just to build a Shenzen in the US" Probably not. Imagine if you went to a flea market, but instead of comic books, hot dogs and used sofas ... you found all sorts of cheap, exotic component parts. Anything. Everything. For 10's of cents each. I'm close with a team making a product, and there are a few 'custom-ish' things they need, like a tiny microcontroller with camera that does some very basic processing as a sub-module. In America - that's a big deal. Hire a team, get some researchers to consult, do some iterations yada yada. In China some guy in a ratty, dirty, poor little corner shop has it! For like 40 cents in units of ONE! There's a massive 'secondary industry' of people making 'not rocket science but not trivial' gadgetry very fast, and on the cheap that is not going to be duplicated anywhere. Add in the tons and tons of cheap labour, with few labour laws. Add in the 10 hours 7 day crazy work weeks. No - Shenzen will not be duplicated. That said - it may be possible to expand the Valley in a more 'automated' 'high end' way. For example: 3D printers that are more amenable to 'short production runs' instead of just prototypes. Low cost / fast turnaround injection moulding. Quick turn around on smaller components. This might make it better for American firms doing quality high tech to churn out faster prototypes, first production runs, and later automated large runs. Doing this would take some champion leaders, maybe Cisco, or Google or a few others setting up shop, proving the way for others to follow.

alvern 8y ago
For example: 3D printers that are more amenable to 'short production runs' instead of just prototypes. Low cost / fast turnaround injection moulding. Quick turn around on smaller components. This already exists in the North American market, I.E. Protolabs
- sonnyblarney 8y ago
  Funny you should mention them I was literally speaking to them this week :) Protolabs is nothing special really. The tech everyone uses for 3D printing, for quick turnaround injection moulding ... it's mostly the same old stuff. 3D printing is only good for a few units, unless what your making is very small and very expensive. Custom moulding is still a big leap - and if there are any iterations (and there will be) it's not cheap. And all of the 'fine' stuff is still done by hand. So I see 3 roadblocks: 1) Rapid iteration and availability of component parts 2) True cheap way to get from 10 to 20000 units i.e. before true mass production 3) 'programmable robot hands' i.e. a very easy to program robot arm/hands that is nimble enough to fold and cut fabric, assemble arbitrary things, wipe/clean surfaces.
  - davidzweig 8y ago
    Well, you can make 100+ high-quality plastic parts a week on a $500 SLA printer (if you can fit at least 5 on the bed, vertically), and the price per part for the material could be below $1/part (the goo costs $70 per liter). At least in theory, that's my plan in the next weeks. PCBs are no issue in smaller quantities, although the setup cost for PCB assembly will add to per unit cost if you are doing less than ~500. Here's an example, although perhaps a little big: https://www.youtube.com/watch?v=RDUn3CoVmuc https://www.youtube.com/watch?v=RDUn3CoVmuc Looks quite saleable to me. Maybe sand blasting would improve the finish. The real plus China having Taobao.. like eBay, but with every imaginable component you could need, and five different Chinese clones of it at 20% of the original component cost, all available at wholesale prices, in quantities of 1 or 100,000. The Shenzhen markets are still active, but shrinking.. you can usually find more variety, faster, cheaper, online. You can hand your BOM over to an agent on Taobao, and he'll get the parts together for you and ship them over.. it might cost 1/4 of the price of Digikey.
    - sonnyblarney 8y ago
      Yes, if the parts are small and don't need much finishing ... that just might fly. Good luck with that I hope it works for you. Off the shelf electronics are good for some things ... not so much for others. Glad to see someone trying it!
- toss1 8y ago
  Yes, but in my experience, ProtoLabs is the exact opposite of everything available for <$1 -- they specialize in rapid turnaround, and they want all the money for that kind of rush. I also found that they aren't all that amenable to finding the most efficient route using stock material. E.g., on one job, I just wanted a stamped or laser-cut hole pattern in stock aluminum, so I specified the stock thickness in that field and commented in the notes what I was seeking. I got back a quote for basically CNC milling the exact part out of billet. Lovely, but completely unnecessary, and it didn't improve chatting with the rep. I would up fabricating the parts out of carbon fiber and CNC cutting the shapes/holes myself (all in-house). So, yes, I'd say that the Shenzen flea-market mentioned above could be really cool, is definitely not here yet, and is sadly unlikely... (Edit: typo, dropped words)
seanmcdirmid 8y ago
> Imagine if you went to a flea market, but instead of comic books, hot dogs and used sofas ... you found all sorts of cheap, exotic component parts. Anything. Everything. For 10's of cents each. Sounds like how Silicon Valley used to be...
- Animats 8y ago
  Yes, when Weird Stuff Warehouse, HSC, and Haltek were still in business. Steve Jobs bought his first oscilloscope at HSC.
nostrademons 8y ago
You'd need to have a way to replace humans with machines and do so both flexibly and at-scale. The big competitive advantage of Shenzhen is people - and not just unskilled labor, but a lot of skilled tradespeople & engineers that can design & build electronic components to spec very quickly. The U.S. just doesn't have the density and skill level for that - the closest you might get is Silicon Valley for software, but even here the population density is < 1/4 of Shenzhen, a relatively small percentage are skilled engineers, and we've somewhat overspecialized on Internet, mobile, & big-data software, making it difficult to innovate in other fields without having an extremely rare combination of skills in one individual. In theory things like 3D printers, pick & place machines, and better CNC millers all controlled by software could help with that, but interest in hardware hacking in the Bay Area has gone down and a lot of the professional electronics manufacturing moved to China.
- sonnyblarney 8y ago
  "but a lot of skilled tradespeople & engineers that can design & build electronic components to spec very quickly" Yes this, is what I failed to mention.
- reaperducer 8y ago
  You'd need to have a way to replace humans with machines and do so both flexibly and at-scale Isn't this something that Elon Musk is working on? I think I read somewhere that one of the reasons his car factory was so expensive to build is because the robots have the capability to do more than just one thing.
  - nostrademons 8y ago
    Hmm, interesting thought: what if the reason Elon Musk is overinvesting in robotics with Tesla is because he believes that the initial Mars colonists will be extremely labor-constrained, and will need robot helpers to survive? Current plans for the BFR and Mars Transporter indicate an initial crew of 12 people, most of whom will likely be highly-trained specialists. If you're building a base around 12 people who are likely biologists, doctors, mechanics, engineers, and rocket scientists, who's gonna clean the toilets and grow the food? Plus robots have the big advantage of not needing food and being able to work outside in the thin, freezing Martian carbon-dioxide atmosphere.
arcticfox 8y ago
> Add in the tons and tons of cheap labour, with few labour laws. > Add in the 10 hours 7 day crazy work weeks. These advantages seem like they'll expire as China becomes more competitive economically. > Imagine if you went to a flea market, but instead of comic books, hot dogs and used sofas ... you found all sorts of cheap, exotic component parts. Anything. Everything. For 10's of cents each. How many types of sensors / components etc. are even conceivable, no matter how exotic? It seems like someone could replicate that in the US with a warehouse and sourcing. These things are extremely cheap in general; it's not like trying to replicate the VCs of Silicon Valley elsewhere. > For example: 3D printers that are more amenable to 'short production runs' instead of just prototypes. Low cost / fast turnaround injection moulding. Quick turn around on smaller components. ... Doing this would take some champion leaders, maybe Cisco, or Google This seems like a great idea; pair some small fabs with a big ol' warehouse stocked with every random component / sensor under the sun, and it seems like it would work out totally fine.
- sonnyblarney 8y ago
  "It seems like someone could replicate that in the US with a warehouse and sourcing. These things are extremely cheap in general;" There are an infinite number of module combinations. They all vary in cost, quality, precision, speed, size, materials, performance, suitability for manufacture, and subtly in whatever it is they do. A commenter above mentioned the vast number of professionals with a ton of know-how in this area, making the 'time to tweak' or 'from nothing to module' and also 'optimized for manufacture' very rapid. There's nothing remotely like it anywhere.

alvern 8y ago

For example: 3D printers that are more amenable to 'short production runs' instead of just prototypes. Low cost / fast turnaround injection moulding. Quick turn around on smaller components. This already exists in the North American market, I.E. Protolabs

sonnyblarney 8y ago
Funny you should mention them I was literally speaking to them this week :) Protolabs is nothing special really. The tech everyone uses for 3D printing, for quick turnaround injection moulding ... it's mostly the same old stuff. 3D printing is only good for a few units, unless what your making is very small and very expensive. Custom moulding is still a big leap - and if there are any iterations (and there will be) it's not cheap. And all of the 'fine' stuff is still done by hand. So I see 3 roadblocks: 1) Rapid iteration and availability of component parts 2) True cheap way to get from 10 to 20000 units i.e. before true mass production 3) 'programmable robot hands' i.e. a very easy to program robot arm/hands that is nimble enough to fold and cut fabric, assemble arbitrary things, wipe/clean surfaces.
- davidzweig 8y ago
  Well, you can make 100+ high-quality plastic parts a week on a $500 SLA printer (if you can fit at least 5 on the bed, vertically), and the price per part for the material could be below $1/part (the goo costs $70 per liter). At least in theory, that's my plan in the next weeks. PCBs are no issue in smaller quantities, although the setup cost for PCB assembly will add to per unit cost if you are doing less than ~500. Here's an example, although perhaps a little big: https://www.youtube.com/watch?v=RDUn3CoVmuc https://www.youtube.com/watch?v=RDUn3CoVmuc Looks quite saleable to me. Maybe sand blasting would improve the finish. The real plus China having Taobao.. like eBay, but with every imaginable component you could need, and five different Chinese clones of it at 20% of the original component cost, all available at wholesale prices, in quantities of 1 or 100,000. The Shenzhen markets are still active, but shrinking.. you can usually find more variety, faster, cheaper, online. You can hand your BOM over to an agent on Taobao, and he'll get the parts together for you and ship them over.. it might cost 1/4 of the price of Digikey.
  - sonnyblarney 8y ago
    Yes, if the parts are small and don't need much finishing ... that just might fly. Good luck with that I hope it works for you. Off the shelf electronics are good for some things ... not so much for others. Glad to see someone trying it!
toss1 8y ago
Yes, but in my experience, ProtoLabs is the exact opposite of everything available for <$1 -- they specialize in rapid turnaround, and they want all the money for that kind of rush. I also found that they aren't all that amenable to finding the most efficient route using stock material. E.g., on one job, I just wanted a stamped or laser-cut hole pattern in stock aluminum, so I specified the stock thickness in that field and commented in the notes what I was seeking. I got back a quote for basically CNC milling the exact part out of billet. Lovely, but completely unnecessary, and it didn't improve chatting with the rep. I would up fabricating the parts out of carbon fiber and CNC cutting the shapes/holes myself (all in-house). So, yes, I'd say that the Shenzen flea-market mentioned above could be really cool, is definitely not here yet, and is sadly unlikely... (Edit: typo, dropped words)

seanmcdirmid 8y ago

> Imagine if you went to a flea market, but instead of comic books, hot dogs and used sofas ... you found all sorts of cheap, exotic component parts. Anything. Everything. For 10's of cents each. Sounds like how Silicon Valley used to be...

Animats 8y ago
Yes, when Weird Stuff Warehouse, HSC, and Haltek were still in business. Steve Jobs bought his first oscilloscope at HSC.

nostrademons 8y ago

You'd need to have a way to replace humans with machines and do so both flexibly and at-scale. The big competitive advantage of Shenzhen is people - and not just unskilled labor, but a lot of skilled tradespeople & engineers that can design & build electronic components to spec very quickly. The U.S. just doesn't have the density and skill level for that - the closest you might get is Silicon Valley for software, but even here the population density is < 1/4 of Shenzhen, a relatively small percentage are skilled engineers, and we've somewhat overspecialized on Internet, mobile, & big-data software, making it difficult to innovate in other fields without having an extremely rare combination of skills in one individual. In theory things like 3D printers, pick & place machines, and better CNC millers all controlled by software could help with that, but interest in hardware hacking in the Bay Area has gone down and a lot of the professional electronics manufacturing moved to China.

sonnyblarney 8y ago
"but a lot of skilled tradespeople & engineers that can design & build electronic components to spec very quickly" Yes this, is what I failed to mention.
reaperducer 8y ago
You'd need to have a way to replace humans with machines and do so both flexibly and at-scale Isn't this something that Elon Musk is working on? I think I read somewhere that one of the reasons his car factory was so expensive to build is because the robots have the capability to do more than just one thing.
- nostrademons 8y ago
  Hmm, interesting thought: what if the reason Elon Musk is overinvesting in robotics with Tesla is because he believes that the initial Mars colonists will be extremely labor-constrained, and will need robot helpers to survive? Current plans for the BFR and Mars Transporter indicate an initial crew of 12 people, most of whom will likely be highly-trained specialists. If you're building a base around 12 people who are likely biologists, doctors, mechanics, engineers, and rocket scientists, who's gonna clean the toilets and grow the food? Plus robots have the big advantage of not needing food and being able to work outside in the thin, freezing Martian carbon-dioxide atmosphere.

arcticfox 8y ago

> Add in the tons and tons of cheap labour, with few labour laws. > Add in the 10 hours 7 day crazy work weeks. These advantages seem like they'll expire as China becomes more competitive economically. > Imagine if you went to a flea market, but instead of comic books, hot dogs and used sofas ... you found all sorts of cheap, exotic component parts. Anything. Everything. For 10's of cents each. How many types of sensors / components etc. are even conceivable, no matter how exotic? It seems like someone could replicate that in the US with a warehouse and sourcing. These things are extremely cheap in general; it's not like trying to replicate the VCs of Silicon Valley elsewhere. > For example: 3D printers that are more amenable to 'short production runs' instead of just prototypes. Low cost / fast turnaround injection moulding. Quick turn around on smaller components. ... Doing this would take some champion leaders, maybe Cisco, or Google This seems like a great idea; pair some small fabs with a big ol' warehouse stocked with every random component / sensor under the sun, and it seems like it would work out totally fine.

sonnyblarney 8y ago
"It seems like someone could replicate that in the US with a warehouse and sourcing. These things are extremely cheap in general;" There are an infinite number of module combinations. They all vary in cost, quality, precision, speed, size, materials, performance, suitability for manufacture, and subtly in whatever it is they do. A commenter above mentioned the vast number of professionals with a ton of know-how in this area, making the 'time to tweak' or 'from nothing to module' and also 'optimized for manufacture' very rapid. There's nothing remotely like it anywhere.