304 North Cardinal St.
Dorchester Center, MA 02124
304 North Cardinal St.
Dorchester Center, MA 02124
Many tech sites, including Engadget, The Verge, ZDNet, Next Reality, and CNET are reporting that Hololens 2 is “officially” shipping as of today as a price of $3,500 for the bare-bones hardware-only version. But before you whip out your wallet, you should be warned that you are probably not special enough to be able to buy one.
At least on the surface, this seems to be an announcement of shipping in name only as it looks like a face-saving release. After all, it has been over 8 months since Hololens 2 was announced and people were beginning to talk.
Microsoft has said for months that they have been shipping a limited number of units to key customers. So it is not clear what has changed with this “announcement” other than making it official that they are shipping limited quantities to very select customers. It not clear whether things like NDA’s have been released, as they would once a product is on sale to the public. We are still not seeing any teardowns, detailed information, or shot through the lens pictures and videos.
As I reported last month, this type of very limited roll-out is consistent with having manufacturing problems. The word on the street is that they are having serious reliability problems with the laser beam scanning engine. It is being spun by marketing that the demand has outstripped the supply, but this can because the supply is very limited.
I have seen several people comment that when they contact Microsoft they get the form letter response such as the one (see left) from Microsoft’s Alex Kipman’s twitter feed. And they are the “lucky ones,” as many others are reporting that Microsoft is not even answering them.
Quoting the letter, “The HoloLens 2 has been released November 7th. As is the case with many products that we develop at Microsoft, we have seeded the market with a small number of HoloLens 2 devices to gather feedback and ensure we are meeting our customer needs. While HoloLens 2 devices are now shipping to the customers that have been selected to give feedback, it is not available for purchase at this time.
We expect HoloLens 2 to remain constrained for several months. We can capture your interest and notify you when units become available to sell. If you have already submitted for pre-order status we would reach out to you once we are able to sell the product.“
Next Reality was reporting earlier in the day that Hololens 2 was “Generally Available” but later printed an update that it was a miscommunication. This kind of problem tends to happen face-saving type press releases that make the announcement sound like there is more to it.
Quoting Next Reality: “Update: While Microsoft PR told Next Reality in no uncertain terms that general availability of the HoloLens 2 would launch today, it appears this was a miscommunication as Microsoft Sales is now telling pre-order customers that while the HoloLens 2 is now officially shipping, they’re still working through their “substantial” backlog of pre-order interest before opening up general availability. If you have pre-ordered by filling out the form on the HoloLens 2 buy page, you’ll still have to wait for Microsoft to contact you to place your order once they’re ready to ship to you. We have asked Microsoft for an estimated timeline for pre-order fulfillment and are awaiting a response. Read on below for our original reporting on today’s launch, including some hands-on time with a production HoloLens 2.”
I suspect the term “General Availability” set off some alarm bells at Microsoft. This would cause people to think that they could actually buy one. It also could release people that have been holding back information under NDA.
In a recent article reviewing the Hololens 2, ITPRO made some critical comments about Hololens 2’s image quality. Quoting them directly below:
Viewing the holograms isn’t akin to viewing images on an LCD display. It’s quite a surreal experience in which the holographs’ borders are defined, yet their colours fuzz and jitter. It’s like viewing images through an old CRT monitor, in one sense, except for the ability to interact via tugs, pulls, pinches and taps. of software servicing the HoloLens 2.
As I have reported, my expectation based on other laser projectors I have seen and the scanning rate specification of the Hololens 2 is that there are going to be problems with the image quality. I have had verbal reports that there is flicker and that there are resolution issues, but this is the first public writing I have seen discussing “colours fuzz and jitter.” Unfortunately, there are zero “through the lens” pictures I can find of Hololens 2. If you know of any, please let me know.
I suspect that this announcement is going to cause Microsoft more problems than it solves. They don’t seem to have the quantities of units available to warrant a big announcement of “shipping.” They will be upsetting the customers that put their name in to buy a unit over 8 months ago. Now there will be a watch to see how long from this announcement until the units are flowing to companies outside a very select few. Frankly, it feels more like a Kickstarter than the way a major company should handle this type of announcement.
I’m curious has anyone outside of a few major companies received a Hololens 2. I’m also curious why there are no “through the lens” pictures. The lack of information suggests that there are very few units out in the wild and there are restrictions on what can be done with them.
According to Microvision, they shipped $1.2M worth of MEMS to their un-named customer (MSFT) in Q3 and expect to ship $2.4-4.4M more in Q4. They had a backlog of $5.5M at the end of Q3 and have since received another order of approx $5M for delivery in Q1-Q2 2020. Without knowing the MEMS cost per Hololens 2 unit, it’s impossible to know exactly how many units that implies, but a WAG of $100 per unit implies approximately 120,000 units over that time frame, not including any additional orders. At $200 per unit, that implies 60,000 units.
Obviously, one should not bet against additional orders or an accelerating ramp. I doubt therefore that supply constraints will last long relative to demand, unless demand proves to be substantially higher than anticipated.
Part of what we don’t know is what “counts” as Microvsion revenue. They use contract manufacturing for just about everything as far as I know. If the whole engine including lasers, optics, ASICs, PCBs goes through them, then the cost per eye (times 2 per unit) could be well over $100/eye and $200/unit. It looks like Microvision is losing money per unit going to Microsoft from their call transcript which suggests they are doing a lot more than licensing. We also don’t know if there are price drop expectations with volume.
Hololens has effectively been obsoleted by the HL2 for well over a year and has had the Osborn Effect (https://en.wikipedia.org/wiki/Osborne_effect) for at least 1 year so there is some pent up demand. HL1 shipped 50K units over 2 years or 25K per year. I could believe that Microsoft would be planning on making about 50K-70K units in 2020 or a bit over 2X what they did with HL1. There are enough variables that the numbers you gave are not unreasonable.
I should add, that if yields or reliability are not solved, all bets are off. The quickest way to lose a lot of money is to ship a product with yield, and worse short lifetime/reliability, problems.
From the CC, it sounded like they yield problems initially which have been resolved.
They said it’s going “smoothly” now which is reflected in the higher anticipated shipments for Q4.
This may be true, but remember the most important part of the conference call is the warning about “The information in today’s conference call includes forward-looking statements, . . . These statements are not guarantees of future performance. Actual results could differ materially from the future results implied or expressed in the forward-looking statements.”
In short, the conference calls goals are generally to put as positive a spin on things as possible. Did they report that their yields were bad before? The word “smoothly” is hardly quantitative or something you could use to say they were lying in court. We don’t know if the “higher than anticipated” (higher than what?) shipments are due to previous problems or if they will happen if some new “unforeseen problem” emerges. Microvision’s track record is not particularly good and thus the stock is trading below $1.
The wheels could always come off, true, but anybody over 20 knows that about everything, so that’s not much to go on. Same for standard boilerplate re. forward looking statements. So what’s left is a low share price after 26 years (and approx. $1B) of development.
MSFT seems cool with that, which speaks volumes.
They have to put the wheels on first before they can come off. The Microvision faithful were saying all the same things about Sony and how did that turn out? Right now you have Microsoft with a “High Tech Lemonade Stand” (like a child running a lemonade stand where they get all the supplies from their mother) where the mother company is spending something like 10X the revenue (at low to no per-unit profit) they are getting from the headsets. Microsoft also seems to be treating Microvision somewhat shabbily, taking all the credit and giving Microvision no credit while stock flounders below $1/share.
There are markets one can target for AR, but the current technology is not ready for it to be a prime-time major market shipping in the millions. This confines AR to markets like “enterprise” that combined are a few hundred thousand per year. You have to divide the total number of eventual users by the lifetime of the product to get the per-year numbers.
I and many others in the industry think the display technology that has the most promise in the future for AR is MicroLEDs which will make electromechanical scanning of lasers look primitive.
I recall Bernard Kress commenting on the attractiveness of combining MEMS scanning with microLEDs. He didn’t expand on it but I remain intrigued. Any comments on the implications of this combination? For example, can microLEDs overcome the vergence-accommodation conflict otherwise?
Continuing the microLED with MEMS mirror scanning point, here’s a Facebook patent application that proposes that: https://patents.google.com/patent/US20170236463A1/en
But, generally, as you have noted elsewhere, given there is no free lunch or Moore’s law for optics, does not even a tiny high resolution microLED panel still require increasingly large optics to expand the FOV without scanning mirrors or some other beam scanning technology? And even if the right balance is struck, won’t the projected image be in focus only within a limited depth of field, in the absence of an additional hardware solution, unlike with scanning mirrors? Simply putting a transparent microLED lens into eyewear won’t work because the image is too close to the eye to be seen, right? So how do microLEDs alone resolve all the challenges of AR, apart from brightness, that other panel approaches have not?
Here’s the original development and components PR from 2017, for what it’s worth.
You won’t find any “through the lens” photos or videos because Microsoft has been _very_ strict about allowing anyone with hands-on time to do so. Those that have been allowed to do so did so under the agreement that their media would be used internally only for purposes such as checking fidelity of colored rendered through the laser display. The reason for this is because it would not be representative of what you actually see through both channels (L and R) when wearing the device.
Source: attended multiple MS developer events with HL2 units present
Thanks for the info.
IMO, the whole “it would not be representative of what you actually see” excuse is just a way of keeping it under wraps, likely because it has problems. If it looked great, then they would be happy to have pictures published via a single eye.
It is tricky to capture the various types of displays since the way they generated the image can “beat” against how a camera works. But you can get a representative picture if you work at it. For some types of displays, it can take a lot of work to make it looks right/good, for others, such as OLEDs and most LCDs, it is very simple.
I was lucky enough to get one a few weeks back for my AR studio 1RIC. Haven’t shared much through the lens footage but there’re nothing holding us back from doing so.
Do you have any stills or video you could share either privately or publicly?
Any update on through the lens footage pls?
Nobody has shown me any through the lens views in still or video. I would expect to start seeing these soon.