[HN Gopher] DARPA and Texas Bet $1.4B on Unique Foundry -3D hete...
       ___________________________________________________________________
        
       DARPA and Texas Bet $1.4B on Unique Foundry -3D heterogeneous
       integration
        
       Author : pseudolus
       Score  : 67 points
       Date   : 2025-11-11 11:33 UTC (11 hours ago)
        
 (HTM) web link (spectrum.ieee.org)
 (TXT) w3m dump (spectrum.ieee.org)
        
       | osnium123 wrote:
       | Skywater Technology offers similar services as well. Competition
       | is good but does the US risk funding too many subscale entities?
       | 
       | https://www.skywatertechnology.com/heterogeneous-integration...
        
         | wbl wrote:
         | The US will fund as many people who will develop this tech as
         | need be. The national champion approach produces a too big to
         | fail lazy company.
        
           | alfiedotwtf wrote:
           | Wouldn't any form of government funding produce a lazy
           | company vs companies standing on their own two feet with the
           | free market acting as drill sergeant?
        
             | AnthonyMouse wrote:
             | Suppose you have to do some R&D in order to make something
             | happen. There is no way to keep it a secret and China isn't
             | going to enforce a patent on it, so if you pay to do the
             | R&D and then have to charge prices high enough to recover
             | it, they undercut you on price and you go out of business.
             | But if the government covers the R&D then you can do
             | domestic production at a competitive price.
             | 
             | Meanwhile the subsidy should be going to every company in
             | the industry so then they still have to compete with each
             | other.
             | 
             | Or to put it a different way, what's really the difference
             | between a subsidy and a tax cut?
        
             | wbl wrote:
             | The funding can get companies spun up and competing.
        
       | oxw wrote:
       | Chip stacking today is limited by thermal management. Can't stack
       | multiple high TDP (e.g. logic) chips together, can only do one +
       | low power chips (e.g. memory). Wonder if this is something DARPA
       | is trying to solve here.
        
       | inshard wrote:
       | If we are thinking about world models and embodied AI as the next
       | big wave, I feel we need to factor the size of this market across
       | military and civilian applications. When I hear Tesla AI4-6 plans
       | for autonomous driving and Optimus, Apple's home robot, and many
       | such initiatives, I'm also thinking of an autonomous military
       | drone arms race and autonomous space exploration and moon
       | colonization. Edge compute will be a key innovation and I feel
       | Apple is gonna be best placed to exploit their power efficient
       | M-series architectures.
        
         | Zigurd wrote:
         | Battlefield robots already exist. They just don't look like you
         | and me. And mostly they fly. That word "embodied"is a broad
         | hint at misunderstanding. Robot surgeons don't stand on
         | humanoid feet next to the operating table, and they don't have
         | humanoid "hands."
         | 
         | Boston Dynamics has had humanoid robots doing parkour for years
         | now and nobody needs a robot to do parkour. This was also a
         | hint.
        
           | inshard wrote:
           | Do we need to distinguish between Atlas, the control theory
           | based humanoid tech stack vs say Optimus that may run on an
           | end-2-end visual reasoning world model? I feel these are two
           | different paradigms. Same for military drones. The next gen
           | prototypes may have launch-and-forget levels of autonomy,
           | however ethically questionable.
        
             | Zigurd wrote:
             | Isn't Optimus the robot made by that guy who thinks it's
             | going to be a "fantastic surgeon?"
        
         | diamond559 wrote:
         | Humanoid robots are a joke to pump stocks. Don't fall for the
         | hype.
        
       | gnarlouse wrote:
       | Because America can no longer guarantee the safety of Taiwan or
       | bear the risk of having TSMC shutdown?
        
       | sailfast wrote:
       | This sounds like a cool initiative, but I am curious where Texas
       | is coming up with the $552M for this investment.
        
         | wat10000 wrote:
         | Looks like it's part of a larger effort to invest in
         | semiconductors: https://gov.texas.gov/business/page/texas-
         | chips-office
         | 
         | The Texas state government's annual budget is well over $300
         | billion, so it's not a huge amount of money for them.
        
         | silisili wrote:
         | I was curious if this was just via tax breaks and whatnot, but
         | apparently not.
         | 
         | If my light research is correct, Texas founded its own CHIPS
         | act in 2023, and shortly after appropriated 552 mil to the UT
         | TIE program. The DARPA money commitment came a year later.
        
       | Animats wrote:
       | This makes sense for chips that integrate sensing and processing
       | on the same chip. The military tends to need things like that.
       | LIDAR sensors need both sensing and fast counting, which often
       | use different semiconductor technologies. The first flash LIDAR
       | chips had an InGaAs sensor chip and a CMOS counting chip, both
       | ball grid arrays, soldered back to back. If different technology
       | layers on the same chip can do the job, that would simplify the
       | parts.
        
         | Tuna-Fish wrote:
         | It doesn't just simplify, it improves capability and saves
         | power because it makes it much easier and cheaper in power to
         | build an extremely wide interface between them.
        
       ___________________________________________________________________
       (page generated 2025-11-11 23:01 UTC)