Bloomberg: To deliver iPhone X on time, Apple crippled Face ID—Update: Apple calls BS

The iPhone X’s singular feature smeared by a single source.

From a report posted early Wednesday by Bloomberg’s Alex Webb and Sam Kim:

As Wall Street analysts and fan blogs watched for signs that the company would stumble, Apple came up with a solution: It quietly told suppliers they could reduce the accuracy of the face-recognition technology to make it easier to manufacture, according to people familiar with the situation…

To boost the number of usable dot projectors and accelerate production, Apple relaxed some of the specifications for Face ID, according to a different person with knowledge of the process. As a result, it took less time to test completed modules, one of the major sticking points, the person said.

It’s not clear how much the new specs will reduce the technology’s efficacy. At the phone’s official unveiling in September, executives boasted that there was a one in a million chance that an interloper could defeat Face ID to unlock a phone. Even downgraded, it will probably still be far more accurate than TouchID, where the odds of someone other than the owner of a phone being able to unlock it are one in 50,000.

My take: Please clarify: Did Apple reduce the accuracy of its face-recognition technology (paragraph 2) or did it relax the requirements of the manufacturers’ testing procedures (paragraph 14)? Might be the same thing, might not.


Bloomberg’s claim that Apple has reduced the accuracy spec for Face ID is completely false and we expect Face ID to be the new gold standard for facial authentication.

UPDATE: I’m not the only observer who sensed something was off about this story.

  • Mike Wuerthele, AppleInsider: “It is not clear where or how Bloomberg sourced the data used for the report, or how accurate it is. Assuming the claims are accurate, it is also not clear is when the decision was made.”
  • Dave Mark, The Loop: “We’ve seen plenty of examples of iPhone (and other Apple product) shortages that lasted months, with demand outstripping supply. I don’t believe Apple would risk the iPhone X reputation by shipping an iPhone X with less-than-effective facial recognition.”
  • Jack Purcher, Patently Apple: “This isn’t just a bad rumor, this is a hit job.”
  • John Gruber, Daring Fireball: “Frankly, I don’t trust anything Bloomberg reports about iPhones any more.” (This is the full Gruber postmortem at better than 2,000 words.)

See also: KGI: Only 2-3 million iPhone Xs available for launch


  1. David Emery said:
    “sources familiar with the situation…”

    This is one of the big problems with journalism. How can we evaluate the truth of a story without insight into the sources? It seems the best we can do is ‘how well do we trust the reporter?’ And as PED points out so often, the track record for most of those reporters covering Apple could be qualified as “piss poor or worse.”

    October 25, 2017
  2. John Kirk said:
    Face ID may or may not be as accurate as advertised. We’ll only know after it ships.

    But this sounds like a complaint about compromises and trade-offs. There has never been, nor will there ever be, an innovative product shipped without the engineers at some point saying: “Ok, we have to stop improving the product and lock in the specifications so we can ship this on time.

    If Face ID is compromised, that’s a bad thing. If Face ID is less than it could have been had Apple waited another six months to ship it, that’s another thing altogether. The question isn’t whether Face ID works as good as it possibly could. The question is whether it works right now.

    October 25, 2017
  3. Fred Stein said:
    I read the Bloomberg article. Agree with John Kirk above that there will always be some “compromising on the road to my horizon”.

    Maybe i’m an iSheep, fanboy, naive, but Tim Cook would have pushed out the delivery date by now if the problems were serious. He knows manufacturing..

    How do you test to 1 in million accuracy of a combination of components? It’s complex. Test criteria change over time as confidence builds. Rumors, even if true at a point in time, may no longer apply.

    So what? iPhone X customers will wait if needed. No long term financial impact. Perhaps short term cost issues if yields don’t meet plan – less than 1% of EPS for one quarter.

    October 25, 2017
  4. Gregg Thurman said:
    “As Wall Street analysts and fan blogs watched for signs that the company would stumble, Apple came up with a solution: It quietly told suppliers they could reduce the accuracy of the face-recognition technology to make it easier to manufacture, according to people familiar with the situation…”

    Yeah, right. The day after AMS provides guidance that exceeded WS consensus (something LUMENTUM is expected to do shortly) Bloomberg comes out with a piece that gets them off the hook for earlier articles that proved false.

    What bothers me most about all these “eyeball” pieces is that Apple has not provided order numbers given to suppliers. Nor has Apple provided any information on delivery dates, ramp speed, etc. Without knowledge of how many widgets Apple has ordered, when they are to be received, etc., there is no way for the media to state that supply/shipments are up or down.

    By the way AMS and LUMENTUM (the single largest) are the 2 largest providers of 3D sensors worldwide (among several) on Apple’s supplier list.

    The media is making assumptions using incomplete data, and reporting it as fact. Journalism professors are rolling in their graves at the breakdown in journalistic standards in pursuit of “eyeballs”. Informed truth is no longer the goal of today’s journalists. Their goals are driven by the advertising sales department.

    Thank you Internet.

    October 25, 2017
  5. George Ewonus said:
    Apple has actually responded and said that the claim that it has reduced the accuracy of Face ID to meed demand is false. It says the probability of a random person unlocking your phone is still 1 in 1 million. Good call David.

    October 25, 2017
  6. James Dearborn said:
    Come on PED, update this Fake News!

    October 25, 2017
  7. Ken Cheng said:
    And, it’s possible that Apple’s original accuracy was intended to be 1 in 1 billion, but that due to engineering difficulty they lowered that to 1 in 1 million, which is what they presented. This could just be the normal push-pull between design engineering and production.

    To be fair, nowhere in the Bloomberg article is the word “crippled” mentioned.

    October 25, 2017
    • David Emery said:
      Not mentioned, but -strongly implied-.

      And PED’s own thoughts, about reduction in piece qualification, is a reasonable explanation.

      October 25, 2017
  8. Fred Stein said:
    1/2 % impact on stock price

    and for how long?

    October 25, 2017
  9. William Kortum said:
    I note that Bloomberg’s Mark Gurman’s name is not on the by line of this Bloomberg article. That may say more than the story itself.

    October 25, 2017
  10. Gianfranco Pedron said:
    It is quite possible that Apple loosened the quality control specifications for specific components somewhat without impacting the accuracy of the final product.

    These are entirely new devices with complex, difficult to predict processes and manufacturing yields so it is likely that initial production was tested with 10 x or much higher safety parameters in order to guarantee promised accuracy of 1M:1. Initial manufacturing process may have output components which where scattered all over the place as far as repeatability of results.

    As the manufacturing process gets refined and tweaked to the point where it is under control and output deviation with respect to target gets repeatably and reliably narrower it would be normal to relax the safety margin initially specified in order to increase yield.

    Again, another hit piece designed to misinform the clueless by people who, in all likeliness, are just as clueless.


    October 26, 2017

Leave a Reply