Apple has a yield problem — and it’s Romeo and Juliet’s fault, it seems
Taipei — Apple has a yield problem. Fixed-income traders can turn away now, there’s nothing for you to see here. The yields I’m talking about are production yields, and Apple is struggling to boost them quickly enough to ship a device that accounts for more than 60% of annual revenue.
Cobbling together various media reports, it seems the key sensors for Apple’s fancy new facial recognition feature are harder to make than first thought, and its suppliers haven’t been able to improve production efficiency quickly enough. That’s a failure for Apple.
Manufacturing can be summarised by a simple equation:
Output rate = capacity * yield
Let’s use baking cookies as an analogy. Your inputs are flour, butter and sugar, and your key equipment is an oven. If you bake 100 cookies, but 75 of them are burnt then your yield is 25%.
If you’re catering for a party and need 80 cookies, then you can either do four rounds of baking, buy more ovens to make larger batches (which takes time and money) or stop burning cookies.
That’s the best solution: not only does the high failure rate smoke up the kitchen, but it wastes buckets of money from discarded ingredients and slows down your output rate.
Early in the development of this year’s iPhone, Apple knew supply of OLED screens was going to be a problem because Samsung Electronics is the dominant maker and it can’t churn out enough.
So Apple made the decision to do two models: the iPhone 8 (without OLED) and iPhone X (with OLED).
Reports indicate that Samsung’s OLED yield is now pretty good, but capacity has been tight this year due to a shortage of equipment.
This brings us to Romeo and Juliet. The star-crossed lovers lend their name to the two parts of the facial recognition system used in the iPhone X, according to the Wall Street Journal: "The trouble with ‘Romeo’ involved complications in assembling the various components into the module."
Apple would have known months ago that these components were difficult to assemble. When test production gets going, usually around June, Apple and its suppliers tend to find a lot of yield problems. That’s normal and Apple’s teams work very closely with everyone in the supply chain to identify issues so they can boost yields heading into mass production a few months later. When they’re confident all obstacles have been overcome — or will be — they sign off and move into volume production.
iPhone’s proportion of Apple sales: 63%
Apple took a big bet in splitting the iPhone into two models this year, and early indicators suggest the lower-spec iPhone 8 isn’t garnering much excitement because many may be holding off for the $999 iPhone X.
Aware of the production challenges, Apple delayed release of the X to ensure it could build up enough inventory to meet opening-weekend demand. The risk is that frustrated customers may get fed up with delays and switch to a rival, or not buy at all.
I don’t have specific knowledge of the Romeo and Juliet difficulty, but the WSJ reports that LG Innotek and Sharp assemble the problematic Romeo modules.
It seems Apple was betting that these two suppliers would be able to boost yields quickly enough before launch date to ensure iPhone X shipments wouldn’t be impacted. That hasn’t yet happened and Innotek’s shares are down 19% from an August 31 high.
With the iPhone X sales party just a month away, Apple needs to act fast to stop burning cookies.
• Culpan is a technology columnist for Bloomberg Gadfly. This column does not necessarily reflect the opinion of Bloomberg and its owners.