I don't even think the current AI companies really have technical moats. It seems like the differentiator is just how much money they have to throw at compute.
Most moats are actually regulatory, ranging from copyright and patents, to anti-competitive regulation, to explicit wealth transfers from taxpayers to the company.
I agree with you that many moats are regulatory, but disagree that most moats are. Networks are powerful moats, customer lockin can work well, brand is a strong moat (at least for a time), speed and culture can function as moats.
Fwiw, I don't think any moat is ultimately perfect. Moats themselves are not meant to permanently stop an attacker - they are delaying tactics at best. Companies that stay ahead continuously re-invent and rebuild their differentiation and defenses.
I don't consider that to be part of a moat because you don't have to cross it before you begin competing with the business. Establishing suppliers and customers is just competing with whatever businesses would also want those deals.
A technical moat is knowledge that you have to have (maybe across multiple employees) before you can even begin to start competing. It's rare to come across any technology that couldn't be cloned by a few smart engineers in 6 months. Much more likely that you aren't allowed to build or sell the technology, or you don't have enough capital, or no one will sell you the parts, or it's illegal to start the business in the first place.
Historically many companies seemed like they had a much deeper moat when the real risk comes out of left field. Blockbuster didn’t die because a competitor suddenly opened a lot of stores instead the basic economics underpinning their business model changed.
Nvidia faces similar risks because they depend on access to world class fabs at reasonable prices. If hypothetically Intel suddenly develops a much better node and then keeps executing Nvidia could be left in the dust with little recourse.
Hm. Examples:
- Amazon? Long road to profitability, but there was fast early growth.
- Space-X, which took a long time to get to a successful launch.
- Someone already mentioned Nvidia.
- Waymo, which took about 15 years from start to real service, and now is doubling in size more than once a year.
- ASML, founded in 1984 and took a long time to become the dominant player in photolithograpy.
- Roblox, tiny for over a decade, then gradually found the right market and really grew.
Who else.
I particularly like Waymo because of aggressively most of the media/talking heads wrote them off as "never going to beat Tesla" because google.
In the end, they took a fundamentally different approach at all levels of the product and won.
Their moat ended up being two fold. One, they were the only tech stack that could handle the scale needed back then. Two, they had a crazy culture of never letting adversaries win and would push daily updates to keep attackers off their customers. Both took a long time to iterate and build towards and nobody thought it was cool while they worked through it.
While, everyone is trying to bast off, we're creating a new category and changing the way people think about sleep.
I don't expect to be able to change the world in a day. This is a movement, and it is going to take time.
I compare it to sunglasses, which existed for centuries as medical aids but didn't become part of everyday life and fashion until they were made "cool" by pilots in WWII.
Throwing rocketfuel might help, but we can build momentum market after market as we go from this somewhat strange activity, monitoring your brain during sleep, and directly increasing slow-wave activity through micro-pulsed auditory stimulation.
The industry is still obsessed with measuring sleep time, we're directly influencing what your brain is doing during that time.
The early momentum of a slingshot takes a bit to get up to speed. Building a comfortable EEG headband for consumers is a challenge, realtime analysis and stimulation take time do develop, among other things.
The momentum begets more momentum. At some point the momentum of the slingshot provides enough force to launch the projectile.
BUT, slingshots don't work forever. Once the projectile is out of the slingshot, it's downward trajectory has already begun.
So I'd prefer an analogy which takes that into account.
If you're curious about our work, check out https://affectablesleep.com
You should read the article before you comment:
"I don’t mean slingshots in the Dennis the Menace sense. Slingshots in the astrophysical sense."
An object that uses an astrophysical slingshot maneuver will continue on its trajectory at the same speed and vector forever (Newton's first law).
