The AI Hardware Dilemma

Was this newsletter forwarded to you? Sign up to get it in your inbox.

Over the past few weeks, new AI-powered hardware has been released to less-than-kind receptions—the Humane Pin was lampooned and the Rabbit R1 was skewered. While some people enjoyed the devices, it is safe to say these were not the launches the companies had hoped for. At the same time, other startups have raised millions in venture capital to build new consumer hardware devices. Investors I know are actively looking to deploy money into the category, and Sam Altman and Jony Ive are in talks to raise up to a billion dollars for a consumer hardware device. 

This disparity raises so many questions: Why are these new devices being received so poorly? What do founders and investors believe they’ll get by betting their careers on this difficult and clearly uphill battle? Will their bets actually work? 

To figure out the answers, I’ve been in the weeds with founders, talking product roadmaps, capital strategies, and levels of excitement. Here’s what I learned.

Why is this hardware’s moment? 

The reasoning behind the rapid new launches and investor bets is simple. It’s AI. 

Many see AI as a technological paradigm shift akin to jumping from personal computers to mobile computing—and there is a chance that a new Apple could be built. With Palo Alto’s favorite fruit-centric company currently enjoying a $2.68 trillion market capitalization, people are feeling like Louis Armstrong did when he touched a trumpet for the first time (quite jazzed).

I’ve previously argued the components of a hardware device can be broken down into three groups:  

1. Silicon: The chips running the computation for the device
2. Interface: How you, as a user, interact with the device 
3. Sensors: The instruments providing data to the software running on a device—cameras, accelerometers, a GPS, a heartbeat sensor, etc. 

New consumer devices are emerging because AI allows you to use the sensors, silicon, and interfaces developed for smartphones in novel ways. AI can take the input of large amounts of ambient data, such as the audio from your conversations and your behavior as you use your computer. Then, AI can output unique insights based on that corpus of data—much more personalized, sensitive, and accurate than what regular software can do today. It can also leverage existing data for new actions, such as following voice commands more closely. Basically, it’ll listen to what you say and bring out smart stuff that you missed.

While my description may be banal, the possibilities are exciting. Smartphones are dominant, but aren’t perfect. Consumer addiction is—at least in my estimation—largely due to the monetization of smartphone app stores and the form factor of the device. If AI can evolve our relationships with devices, it’s a meaningful change. An AI ambiently crunching data and performing tasks without the distraction of a screen is net good for humanity. And, as a happy capitalistic coincidence, it would probably make the inventor of said device the owner of several large yachts. 

So if the promise is so wonderful, why is the category so challenging?

The problem of the iPhone

The answer is, once again, simple: The iPhone is too damn good. 

If I asked you to visualize what the iPhone disrupted, you’d probably think of an image like this—a brick cell phone, a digital camera, a GPS.