Yearly Focus

Technology Predictions – Summer 2024

This post is the first in a new series where I talk about technologies and make guesses about how they’ll deliver value in the future. I’ve actually been doing this privately for a few years now and recently decided I should share my thoughts publicly, as I’ll surely learn more that way – both from the responses as well as from the extra thought I’ll put into it before I hit publish.

One of my goals in life is to be a great technologist – an expert at identifying technology that can deliver value to consumers and businesses. A lot of people try to be great technologist, but many of them miss. Sometimes it is poor understanding the technology, sometimes it is delusion or denial, sometimes it is lack of execution. I hope this blog series will help me with the former two.

With the hope of improving over time – here we go…

Tech I think is undervalued

  • Spacial compute – people seem very hesitant about it, but I think there is a lot of value here. Being able to overlay information on top of what I view is 10X more valuable than forcing me to switch context and focus by looking over at a screen. Pass through has to work amazingly well for this though and we aren’t there yet. Virtual rectangular screens are a great next step. Ultimately we’ll need a platform that opens up APIs for developers to render things not on a known rectangular screen, but on an unknown-at-the-time-of-programming image of the world. Apple is in a great place to start to do the discovery for what those APIs should be. I’ll confess I didn’t buy an Apple Vision Pro – I couldn’t think of enough of a valuable enough use case for me (I already have large rectangles) and didn’t want a $4k paperweight. I’m glad that some people are funding that innovation though and look forward to the day when the capabilities expand enough and developers put something useful enough together so that I have a need to get one.
  • Networking – low latency, highly secure and reliable – it seems silly to add this, networking isn’t exactly the hot tech it was 25 years ago. I think it is being undervalued though. Perhaps people are still burned out after the 5G hype wave. Reliable data transfer is required to power lots of current and future tech and the existing end-to-end-error-rates simply aren’t low enough for most people and many use cases. Perhaps this is a field of slow, steady and quiet progress, but I can’t help but imagine there is some step change innovation in our future.
  • Brain interfaces – I’m quite impressed by recent research in capturing inputs directly from the electrical signals inside of a human and using them to complete a task like move a robotic arm, create a picture of the image they were looking at or enter text. There are obvious use cases for those missing a limb, but I think there is a lot more value there is removing abstraction layers. We talk about the value of Gen-AI as using English as the interface, removing the need to translate through the proxy of code. But English is a proxy for thoughts which has limitations – words that don’t yet exist, not everyone speaks it – and introduces latency. I don’t think the final form is Facebook’s bracelet that lets you type on anything, but I think that is a good step. Today to generate an image I need to go from a thought to an English phrase that I type using my fingers onto a keyboard which turns it into electric signals that go to the computer and eventually to a GenAI system that can process the info and render pixels. Skipping the fingers touching the keyboard part is cool , but ultimately we should be able to go from mental image to pixel image without needing the words or typing motions. Perhaps we don’t need external compute either.

Tech I think is appropriately valued

  • Small EVs in developing nations – there will be more EVs than ICEs sometime in the next 10-20 years. I don’t think there will be more EVs registered in the USA than ICEs any time in the next 20 though. Most of the EV adoption is going to come from places in the world that have more bikes than cars today. I think there is a ton of value in using EVs as the framework to power transportation and letting the form factor match the job. The Cyber Truck is not a very practical EV – 90-97% of the energy used is to transport the vehicle and only 3-10% the load, for most of the miles driven. A small EV Tuk Tuk seems like something much more practical and affordable for 99% of the world. When you travel the world you see that a scooter can, and often is, used as a 4 person family vehicle. I think more of the next billion vehicles added will look like that, or slightly bigger than American-style tank-cars.
  • IoT – Does anyone even use that word anymore? I think manufacturers have moved on to putting AI in every high-end device as their vector to capture more margin from low-price-elasticity buyers. Meanwhile IoT is quietly doing some things very well. I’m yet to be convinced a fully integrated smart home can work for anyone who wants to spend less than a full time job maintaining it. There are too many manufacturers that seem to not play well together, sometimes even with themselves. (Shoot, Google can’t even automatically connect THEIR speakers with THEIR Chomrecast – I still have to manually link them before each movie I stream.) But my sprinkler system reliably uses wifi to not water the lawn on days where it is going to rain – which was well worth the $100 device that plugged into the middle of an existing set of infrastructure. Sure, I need a dedicated standalone app – but that is fine for that job and it doesn’t need to know anything about what my dishwasher is doing.
  • On-device generative AI – I almost called this under-valued, but I’ve seen more people talking about this and how Apple and Google will need to do it well. The limitations on data transfer and the cost of compute will make it impossible to scale generative AI with today’s model of running it centrally. We need smaller models that can run on device to perform most jobs. The good news is this seems completely possible and already in the works.
  • Autonomous Driving – Can we get a car to drive? Yes, in many places. Will we let cars drive? Yes, in some places. Will it be more affordable or practical to have the car drive itself? I’m not so sure. I’m not convinced it is worth it to run a taxi business that requires so much CapEx – perhaps if you have billions you need to do something with and are ok with a 2+ year breakeven, but that is a big bet to take. Will enough people replaces their cars with self-driving. It seems more likely that assisted features will continue to have lower costs and get bundled in similar price-points as today than that we’ll see a price-point leap for a new capability. Freight seems like an obvious win, I’m surprised I haven’t seen more here and suspect it has more to do with existing systems than technology. All in all I think folks are valuing this space appropriately – there is money to be made there but big battles to be won and at this point it is mostly regulatory and cost efficiency problem.

Tech I think is overvalued

  • EVs in the USA – I’m can’t see that the change will be worth it for enough people outside of those that feel morally inclined. The gasoline infrastructure is too good , the electrical grid is too fragile to replace it, the form factor of vehicles isn’t changing enough to take advantage of the benefits of the new technology. At the end of the day, the American consumer is too tied to a very large car and transporting thousands of pounds of metal is never going be be efficient.
  • Food delivery via humans – maybe this isn’t even tech. Ordering from an app is a great way to turn a $9 menu item into a $30 one. What is strange is this problem was solved 50 years ago with pizza delivery and it only added a few dollars of tip to the cost. Now we’ve distributed it and lost efficiency, though I will admit we gained some extra choice along the way. I suspect we’ll see a big regression here. There are specialized cases where higher density, predictable paths and more limited options might make this more automateable – college campus food delivery of a more limited menu comes to mind. Ultimately taking a problem and adding more people that want to get paid into the middle of the solution without fundamentally changing the physics or finding new efficiencies, seems like a poor way to make money.
  • AGI – I don’t think we’re anywhere close. Will LLMs eventually be shown as a stepping stone along the path that helped fund development in infrastructure and research that eventually led to AGI. Probably. Does that mean we’re a model or two away from it. I really don’t think so. The more I learn about how AI works, the more impressed I am with the human brain. I suspect we might eventually reach a point where we realize we have a really powerful, highly efficient, fairly resilient computing device right on top of our heads. If we would just spend some more time installing the right software in it.

Tech I believe in, but think we’re far from

  • Metaverse/Digital Life – I believe in this long term but think we’re a long way from it. Oddly enough I think we have almost all of the technology we need to make a valuable experience, we don’t have the right incentives. Music streaming was enabled by music piracy. There is no way the labels would have willingly taken on the transformation, but they were sort of forced into it and in the end we’re seeing they’ve shifted their revenue and are thriving. I suspect we’ll see a similar pattern with the metaverse. We have too many companies that favor closed instead of open and who have hopes of consolidation and control. If enough of them face enough of a struggle, they might accept open and we might be able to have something cool. Ultimately a person’s willingness to invest in a digital life is capped by their knowledge of the cyclical nature of digital worlds. I think we’d need to see some sort of transformation for the order of magnitude people were willing to spend there go up dramatically.
  • NFTs – it was never going to be images. Reliable records of ownership that span beyond one company/entity are critical for a world that is increasingly digital though. I suspect this relates to the above item. Perhaps neither is a precursor, but they are co-dependent.
  • 3D printing – It is too time intensive, and fragile to be more than a hobby for people that like debugging. This space needs a few order of magnitude of performance and usability improvement – as well as a smaller form factor – to be more widely adopted. I spend a few hundred dollars a year getting small plastic pieces shipped to my house from China though – it seems inevitable that eventually the manufacture can be more localized.

Tech I think is close to disruption

  • Charging cables – data moves via wifi, electricity through magnetic induction. Having 50 different cable types when you have those two standards seems doomed. Perhaps it will take a while to get enough adoption that it is the only option. I bet we see a version of the iPhone that is portless within the next 3 years though.

Tech that is right but in the wrong form

  • Generative-AI – chat is the wrong interface. Products built to solve use cases where 90-95% accuracy is acceptable have a big opportunity but they need a different interface. I suspect we’re sitting at a similar point to where the personal computer was right before the GUI was introduced. I suspect entertainment is the killer app. 95% right is great for music, images or something I’m doing for enjoyment and can skip if it doesn’t feel right. 95% isn’t great for decision making. What if we get to 99% What about 99.9% That could be interesting – but the trajectory seems to be decelerating as we approach it – I’m not sure we’ll get there until there is another step function change.