13 ms·

In what world does it take 22 billion dollars to develop an AR goggle program. Its not a mystery why we have out of control spending and certain groups lobby a

In what world does it take 22 billion dollars to develop an AR goggle program. Its not a mystery why we have out of control spending and certain groups lobby and get wonderfully rich off the governments incompetence and corruption.

ydkme 5y ago
Two thoughts: - I'd be very surprised if Apple releases their AR headset without a similar-size (70% or more) investment over the preceding decade. RDT&E for environment-aware devices is incredibly expensive. - Defense funding and procurement amounts often shock those not in the industry, but the truth is that the entire market for most of what the DoD funds is...the DoD. Sure, a lot of the components and IP can be reused in the commercial sector, but for most programs the DoD has to fund both the initial R&D and the unit sales, because the market is so much smaller and device sales are heavily restricted (e.g. ITAR). The DoD also has MUCH broader and stricter requirements for security, safety, reliability, and compliance with a daunting array of standards and processes than the commercial industry. To be clear, this is on net a good thing for the DoD and by extension the rest of the USA, but it does increase development costs. Certainly there is profit being made by Microsoft here, that's kind of the point of partnering with the DoD. But the intuitive comparison of "DoD programs should cost the same as developing a commercial equivalent" that many of us make is fundamentally apples and oranges in both the product development and the economics.
onlyrealcuzzo 5y ago
In a world where engineers at MSFT make $300k+ - that pays ~70k engineers for a year. There are nowhere near 70k engineers working on this - I'd be surprised if even 7k - but this has been in development for years, and will be in development for a few more. Depending on how many headsets they planned to deliver - I wouldn't be surprised if they each cost more than $1k - there are like 60k infantry soldiers - so that could easily cost >$60M. I'd imagine they spent >$100M lobbying on this and Jedi. I can't see how this would cost anywhere near $22Bn - but military contractors are known to have profit margins >30%. I'm obviously missing a lot of things - but I wouldn't be surprised if this cost MSFT >$10Bn.
- texasbigdata 5y ago
  The headsets currently retail for $6k each and they are consumer not military specs.
- wespiser_2018 5y ago
  The one piece of this that might end up costing a ton of money is the testing using live soldiers. Each of those deployments, along with all the heavy equipment is a lot of money. I saw they already have 80k hours of soldier feedback, so in part, that's where some of the money goes. The rest, of course, is figuring out how to make the HoloLens light and reliable, which no one has ever done. Idk, this could be a huge money pit. The Army has tried this idea before and failed. Now might be the time, or this could be another JSF spending fiasco. Either way, MSFT is doing well.
- cryptica 5y ago
  This is job creation porn. These engineers don't deserve any public money, let alone to prop up their $300K per year salaries. Microsoft could get better engineers for cheaper. Including within the US. The problem is that this isn't about engineering. It's a scheme.
  - onlyrealcuzzo 5y ago
    If MSFT could get better engineers for cheaper - they would. They're a business... Not a charity to engineers.
    - cryptica 5y ago
      This is BS. I know how the interview process works. All these big corporations optimize their recruitment processes around pedigree (matched to the curriculum of top universities). They don't look at track record at all; when track record is a far better indication of future performance. That's why their engineers are so incompetent and it costs so much money for them to produce anything. They have become so incompetent that they don't even have the talent necessary to know what talent looks like when recruiting. Corporate engineers will never hire other engineers who are better than them (for political reasons), so the engineering talent in these companies has been degrading for years as the bureaucracy and internal politics has increased.
wespiser_2018 5y ago
A maze or requirement, bureaucracy, and interoperability with existing and TBD systems? Software is expensive to build. Reliable software all that much more. If you consider an alternative, just make an app and stick phone on a rifle, sure, that will work for 90% of the stuff you want to do, but it would break and that's bad. A true "heads up display" that you can bet your life on, anywhere on earth, at any time, is just a truly massive undertaking.
- dralley 5y ago
  I'm sorry but even taking into account the clusterf*** of government procurement this is a truly obscene price tag. Even by government standards, this is ridiculous. It's half as much as the development of the entire flipping Space Shuttle. It's more than Apple's entire yearly R&D budget. It's more than the James Webb Space Telescope and the entire Hubble program (including operations and the repairs that were needed post-launch) combined. It's roughly the cost of 8 Mars rovers of the same type as Curiosity / Perseverance. This is worse, even, than the Joint Strike Fighter program, which at least started out reasonably with only a pair of 1 billion dollar grants handed out to Boeing and Lockheed to develop two functional prototype fighter aircraft (never mind how the costs spiraled later). Setting aside the question of how much it should or shouldn't cost to develop such a product, it's absolute insanity to do it with a single massive project. The DoD loves to claim in public they're trying to be more agile, and this is the exact opposite of that. This ought to have been done $100 million (or less) at a time with limited initial scope and incrementally expanding goals. Don't make excuses for this. There is no possible justification.
  - whb07 5y ago
    I generally agree with you, but 2021 USD is literally trash vs 1980 USD. For reference, an ounce of gold in the peak at 1980 (which was a long while peak btw) was $800ish, an ounce of gold today is $1800. So more than 2x. So $22B today is more like $10B in 1980.
    - dralley 5y ago
      I tried to use inflation adjusted costs. The "original" development cost of the Space Shuttle was, actually, about 10 billion dollars. Which is around 49 billion in 2020 dollars, adjusted for inflation (gold is a poor comparison, it's closer to 5x than 2x).
ahepp 5y ago
It was a competitive bidding process, no? I don't mean to suggest there aren't huge problems with the army's acquisition process, but it seems no one wanted to make them for less.
- trident5000 5y ago
  I dont think the list of contractors the Army considers is very deep; its not competitive in a true sense. That may be part of the problem.
  - ahepp 5y ago
    I think a lot of what stifles competition is the military itself. They aren't a good customer. They have lots of broken legacy systems. There are serious reputational risks. I don't really mean to suggest 22b is too low a cash prize, but I imagine it takes a lot to overcome the barriers that have been wilfully erected. Much less the ones discreetly erected by incumbent military contractors.
jakey_bakey 5y ago
Playing devils advocate here, Magic Leap has managed to burn through $3.5 billion and has made no achievements of note - instead of solving the two primary optics challenges of AR, they allocated their capital to make blacks look slightly blacker.
m0zg 5y ago
I worked with a Federal Contractor once (you, as a rule, have to go through a middleman to work with US Govt, and that's where most of the profit ends up). Their requirement was that we use their hardware in their datacenter to do our work and never move data out of there. So they asked us for configs. I priced it out, it came to something like $120K in total, sent them the specs. They tore it up and bought $1.2M worth of hardware instead. Insane CPUs, maxed out RAM, Infiniband, you name it. We didn't need any of that. Literally 10x the price. Ever since then I have extra bad taste in my mouth when filing my taxes, and I know for a fact that merely passing money through the government makes most of it disappear into a black hole.
- hooande 5y ago
  The tax dollars didn't completely disappear. They were used to pay you and hundreds of thousands of others
  - m0zg 5y ago
    This is the classical "broken window fallacy", for which there are multiple convincing refutations. I'm sure you can find them yourself. But besides that, wouldn't you want most of your tax money spent on the things it is purportedly spent on (setting aside even whether spending it on those things makes sense in the first place, which is something we the people have zero control over), without leaving most of it in the pockets of whoever owns the federal contractor?
greenreptar 5y ago
it's "up to 22 billion over 10 years." So $2b/year once things get moving. They're still in testing, so likely less than that figure.

ydkme 5y ago

Two thoughts: - I'd be very surprised if Apple releases their AR headset without a similar-size (70% or more) investment over the preceding decade. RDT&E for environment-aware devices is incredibly expensive. - Defense funding and procurement amounts often shock those not in the industry, but the truth is that the entire market for most of what the DoD funds is...the DoD. Sure, a lot of the components and IP can be reused in the commercial sector, but for most programs the DoD has to fund both the initial R&D and the unit sales, because the market is so much smaller and device sales are heavily restricted (e.g. ITAR). The DoD also has MUCH broader and stricter requirements for security, safety, reliability, and compliance with a daunting array of standards and processes than the commercial industry. To be clear, this is on net a good thing for the DoD and by extension the rest of the USA, but it does increase development costs. Certainly there is profit being made by Microsoft here, that's kind of the point of partnering with the DoD. But the intuitive comparison of "DoD programs should cost the same as developing a commercial equivalent" that many of us make is fundamentally apples and oranges in both the product development and the economics.

onlyrealcuzzo 5y ago

In a world where engineers at MSFT make $300k+ - that pays ~70k engineers for a year. There are nowhere near 70k engineers working on this - I'd be surprised if even 7k - but this has been in development for years, and will be in development for a few more. Depending on how many headsets they planned to deliver - I wouldn't be surprised if they each cost more than $1k - there are like 60k infantry soldiers - so that could easily cost >$60M. I'd imagine they spent >$100M lobbying on this and Jedi. I can't see how this would cost anywhere near $22Bn - but military contractors are known to have profit margins >30%. I'm obviously missing a lot of things - but I wouldn't be surprised if this cost MSFT >$10Bn.

texasbigdata 5y ago
The headsets currently retail for $6k each and they are consumer not military specs.
wespiser_2018 5y ago
The one piece of this that might end up costing a ton of money is the testing using live soldiers. Each of those deployments, along with all the heavy equipment is a lot of money. I saw they already have 80k hours of soldier feedback, so in part, that's where some of the money goes. The rest, of course, is figuring out how to make the HoloLens light and reliable, which no one has ever done. Idk, this could be a huge money pit. The Army has tried this idea before and failed. Now might be the time, or this could be another JSF spending fiasco. Either way, MSFT is doing well.
cryptica 5y ago
This is job creation porn. These engineers don't deserve any public money, let alone to prop up their $300K per year salaries. Microsoft could get better engineers for cheaper. Including within the US. The problem is that this isn't about engineering. It's a scheme.
- onlyrealcuzzo 5y ago
  If MSFT could get better engineers for cheaper - they would. They're a business... Not a charity to engineers.
  - cryptica 5y ago
    This is BS. I know how the interview process works. All these big corporations optimize their recruitment processes around pedigree (matched to the curriculum of top universities). They don't look at track record at all; when track record is a far better indication of future performance. That's why their engineers are so incompetent and it costs so much money for them to produce anything. They have become so incompetent that they don't even have the talent necessary to know what talent looks like when recruiting. Corporate engineers will never hire other engineers who are better than them (for political reasons), so the engineering talent in these companies has been degrading for years as the bureaucracy and internal politics has increased.

wespiser_2018 5y ago

A maze or requirement, bureaucracy, and interoperability with existing and TBD systems? Software is expensive to build. Reliable software all that much more. If you consider an alternative, just make an app and stick phone on a rifle, sure, that will work for 90% of the stuff you want to do, but it would break and that's bad. A true "heads up display" that you can bet your life on, anywhere on earth, at any time, is just a truly massive undertaking.

dralley 5y ago
I'm sorry but even taking into account the clusterf*** of government procurement this is a truly obscene price tag. Even by government standards, this is ridiculous. It's half as much as the development of the entire flipping Space Shuttle. It's more than Apple's entire yearly R&D budget. It's more than the James Webb Space Telescope and the entire Hubble program (including operations and the repairs that were needed post-launch) combined. It's roughly the cost of 8 Mars rovers of the same type as Curiosity / Perseverance. This is worse, even, than the Joint Strike Fighter program, which at least started out reasonably with only a pair of 1 billion dollar grants handed out to Boeing and Lockheed to develop two functional prototype fighter aircraft (never mind how the costs spiraled later). Setting aside the question of how much it should or shouldn't cost to develop such a product, it's absolute insanity to do it with a single massive project. The DoD loves to claim in public they're trying to be more agile, and this is the exact opposite of that. This ought to have been done $100 million (or less) at a time with limited initial scope and incrementally expanding goals. Don't make excuses for this. There is no possible justification.
- whb07 5y ago
  I generally agree with you, but 2021 USD is literally trash vs 1980 USD. For reference, an ounce of gold in the peak at 1980 (which was a long while peak btw) was $800ish, an ounce of gold today is $1800. So more than 2x. So $22B today is more like $10B in 1980.
  - dralley 5y ago
    I tried to use inflation adjusted costs. The "original" development cost of the Space Shuttle was, actually, about 10 billion dollars. Which is around 49 billion in 2020 dollars, adjusted for inflation (gold is a poor comparison, it's closer to 5x than 2x).

ahepp 5y ago

It was a competitive bidding process, no? I don't mean to suggest there aren't huge problems with the army's acquisition process, but it seems no one wanted to make them for less.

trident5000 5y ago
I dont think the list of contractors the Army considers is very deep; its not competitive in a true sense. That may be part of the problem.
- ahepp 5y ago
  I think a lot of what stifles competition is the military itself. They aren't a good customer. They have lots of broken legacy systems. There are serious reputational risks. I don't really mean to suggest 22b is too low a cash prize, but I imagine it takes a lot to overcome the barriers that have been wilfully erected. Much less the ones discreetly erected by incumbent military contractors.

jakey_bakey 5y ago

Playing devils advocate here, Magic Leap has managed to burn through $3.5 billion and has made no achievements of note - instead of solving the two primary optics challenges of AR, they allocated their capital to make blacks look slightly blacker.

m0zg 5y ago

I worked with a Federal Contractor once (you, as a rule, have to go through a middleman to work with US Govt, and that's where most of the profit ends up). Their requirement was that we use their hardware in their datacenter to do our work and never move data out of there. So they asked us for configs. I priced it out, it came to something like $120K in total, sent them the specs. They tore it up and bought $1.2M worth of hardware instead. Insane CPUs, maxed out RAM, Infiniband, you name it. We didn't need any of that. Literally 10x the price. Ever since then I have extra bad taste in my mouth when filing my taxes, and I know for a fact that merely passing money through the government makes most of it disappear into a black hole.

hooande 5y ago
The tax dollars didn't completely disappear. They were used to pay you and hundreds of thousands of others
- m0zg 5y ago
  This is the classical "broken window fallacy", for which there are multiple convincing refutations. I'm sure you can find them yourself. But besides that, wouldn't you want most of your tax money spent on the things it is purportedly spent on (setting aside even whether spending it on those things makes sense in the first place, which is something we the people have zero control over), without leaving most of it in the pockets of whoever owns the federal contractor?

greenreptar 5y ago

it's "up to 22 billion over 10 years." So $2b/year once things get moving. They're still in testing, so likely less than that figure.