It's time for Apple to look at owning factories again

Toshiba plant to make LCD panels for Apple: report | Reuters.

Apple will invest in a portion of the investment for the factory, the Nikkei said.

This is one of the more interesting news items I’ve seen for a while.

[UPDATE] Toshiba denies rumor Apple will invest in mobile display subsidiary

There was a time when Apple designed and owned factories. From an interview with John Sculley:

That went all the way through to the systems when he built the Macintosh factory. It was supposed to be the first automated factory but what it really was a final assembly and test factory with a  pick-to-pack robotic automation. It is not as novel today as it was 25 years ago, but I can remember when the CEO of General Motors along with  Ross Perot came out just to look at the  Macintosh factory. All we were doing was final assembly and test but it was done so beautifully. It was as well thought through in design as a factory–a lights out factory requiring many people–as the products were.

During the 90s the manufacturing function moved outside Apple’s ownership umbrella. It was more economical, more flexible and more scaleable to outsource manufacturing to Asia. The time has come to rethink this.

For the past few decades outsourced manufacturing has grown in scale and scope as prices fell and quality increased. However there are signs that this global web of suppliers and assemblers is reaching a point of disruptive change.

Press attention has focused on labor practices, pay structures and the enormity of factory towns where literally millions of workers are employed and housed. As is often the case press attention only hints at some causality and as such misses the point. Journalists looking for a story about child labor or suicidal workers might serve us better by putting the spotlight on the fact that factory towns are not competitive.

centralized Manufacturing is not good enough for the coming mobile computing disruption

The real problem with Chinese factory towns is that they are not flexible enough. By concentrating facilities and workers, electronics manufacturers have improved on measures of scale and quality but are still only able to ramp up and down on individual products once or twice per year. This does not meet the need to increase responsiveness to markets and the pace of development and design. It also reaches some critical overweight mass. Capacity cannot double or triple again to meet the need of six billion consumers. There are signs that the business model must change.

Typically a product is born in Product Marketing, progresses through Design, R&D and then to Manufacturing and finally to Sales. Each of these stages of development has to be re-started for each new product, often in a pipeline fashion. However, some of these stages are increasing their cycle times while others are at their limits. This limits the throughput of the whole chain. If, for example, R&D can put out two product designs in the space it takes for one Sales cycle then they need to be tasked with different product while they wait for “capacity” in Sales to catch up.

My contention is that the bottleneck in device product development is now manufacturing. The evidence is that Apple (among many others) cannot meet demand. Sometimes it’s a lack of components, but often there is a forecast error and capacity is simply not allocated. What I also see is that this bottleneck is not solvable with the current value chain architecture. When there are 5 billion consumers of platform devices, centralized factory town manufacturing does not match the cycle time of either design or of consumption.

The answer to low capacity is more investment in more lines of production, more workers and more suppliers. However the problem is that this takes time. The window of opportunity for a product is less than six months.

Consider the iPhone product cycle. A new variant goes into development 18 months to two years before it’s launched. It’s ready for manufacturing a few months before launch. The R&D team is off making the next version long before the product hits any shelves. The cycle time for R&D changes unpredictably. Sometimes it slips, sometimes it may be shorter because the changes are minor. Perhaps marketing sees the need to react to changing market conditions. However, manufacturing plods along at an inflexible rate.

How can it be fixed?

I’m not an expert in this area, but as a student of disruption I see signs of opportunity to re-engineer the manufacturing business model. The patterns from history are plain to see: Centralized systems get broken down and re-aligned with new bases of competition. Value chains disintegrate and re-integrate as profit algorithms change.

The new algorithm says you need to ramp production quickly (up and down) and to increase “product turns” from one to two each year. Each ramp needs to be even bigger than they are now. Maybe by a factor of two. Products should be built closer to where they are consumed to drive costs and delays out of transportation and tariffs. Carbon footprints need to be reduced.

Sounds daunting. But the rewards could be enormous. The business model may need re-definition. By owning the means of production, the value added can be shifted to another point (i.e. production can be run at a loss in exchange for higher margins elsewhere.)

All this requires capital. Lots of it. Apple has it. I see no better use for it than to re-define production as a synchronized development module.

It’s time to look at owning factories again.