Infrastructure owners with access to the cheapest energy will be the long run winners in AI.
According to Google (or someone at Google) no organization has moat on AI/LLM [1]. But that does not mean that it is not hugely profitable providing it as SaaS even you don't own the model or Model as a Service (MaaS). The extreme example is Amazon providing MongoDB API and services. Sure they have their own proprietary DynamoDB but for the most people scale up MongoDB is more than suffice. Regardless brand or type of databases being used, you paid tons of money to Amazon anyway to be at scale.
Not everyone has the resource to host a SOTA AI model. On top of tangible data-intensive resources, they are other intangible considerations. Just think how many company or people host their own email server now although the resources needed are far less than hosting an AI/LLM model?
Google came up with the game changing transformer at its backyard and OpenAI temporarily stole the show with the well executed RLHF based system of ChatGPT. Now the paid users are swinging back to Google with its arguably more superior offering. Even Google now put AI summary as its top most search return results for free to all, higher than its paid advertisement clients.
[1]Google “We have no moat, and neither does OpenAI”:
Improvements seem incremental and smaller. For all I care, I could still happily use sonnet 3.5.
Quality was rarely the reason open source lagged in certain domains. Most of the time, open source solutions were technically superior. What actually hurt open source were structural forces, distribution advantages, and enterprise biases.
One could make an argument that open source solutions often lacked good UX historically, although that has changed drastically the past 20 years.
Yes, Ardour. It’s no more a toy than KiCad or Blender.
Why would a company use an expensive proprietary model on Vertex AI, for example, when they could use an open-source one on Vertex AI that is just as reliable for a fraction of the cost?
I think you are getting at the idea of branding, but branding is different from security or reliability.
and then we have to look at responsiveness, if the two models are qualitatively in the same ballpark, which one runs faster?
Which is an interesting point in favour of the human employee, as you can only consolidate scape goats so far up the chain before saying "It was AIs fault" just looks like negligence.
Better (UX / ease of use)
Lock in (walled garden type thing)
Trust (If an AI is gonna have the level of insight into your personal data and control over your life, a lot of people will prefer to use a household name)
Not Google, and not Amazon. Microsoft is a maybe.
In the developed world. I'm not sure about globally.
All they have to do is completely switch the google homepage to gemini one day.
ChatGPT feels better to use, has the best implementation of memory, and is the best at learning your preferences for the style and detail of answers.
So a couple of things. There are going to be a handful of companies in the world with the infrastructure footprint and engineering org capable of running LLMs efficiently and at scale. You are never going to be able to run open models in your own infra in a way that is cost competitive with using their API.
Competition _between_ the largest AI companies _will_ drive API prices to essentially 0 profit margin, but none of those companies will care because they aren't primarily going to make money by selling the LLM API -- your usage of their API just subsidizes their infrastructure costs, and they'll use that infra to build products like chat gpt and claude, etc. Those products are their moat and will be where 90% of their profit comes from.
I am not sure why everyone is so obsessed with "moats" anyway. Why does gmail have so many users? Anybody can build an email app. For the same reason that people stick with gmail, people are going to stick with chatgpt. It's being integrated into every aspect of their lives. The switching costs for people are going to be immense.
Google would love a cheap hq model on its surfaces. That just helps Google.
All the facts say otherwise to your thoughts here.
Given how often new models come out, it’s also easier to update an API call than constantly deploying model upgrades.
But in the long run, I hope open source wins out.
They won't. Actually, even if open models aren't competitive, they still won't. Hasn't this been clear since a while already?
There's no moat in models, investments in pure models has only been to chase AGI, all other investment (the majority, from Google, Amazon, etc.) has been on products using LLMs, not models themselves.
This is not like the gold rush where the ones who made good money were the ones selling shovels, it's another kind of gold rush where you make money selling shovels but the gold itself is actually worthless.
For a sufficiently low cost to orbit that may well be found in space, giving Musk a rather large lead. By his posts he's currently obsessed with building AI satellite factories on the moon, the better to climb the Kardashev scale.
Earth based computers benefit from the existence of an atmosphere to pull cold air in from and send hot air out to.
A space data center would need to entirely rely on city sized heat sink fins.
And aluminum is abundant in the lunar crust.
If launch costs are cheap enough, you can bring aluminum up from earth.
But once your in-space economy is developed enough, you might want to tap the moon or asteroids for resources.
That was fun to write but also I am generally on board with humanity pushing robotics further into space.
I don't think an orbital AI datacentre makes much sense as your chips will be obsolete so quickly that the capex getting it all up there will be better spent on buying the next chips to deploy on earth.
Radiative cooling is really annoying, but it's also an engineering problem with a straightforward solution, if mass-in-orbit becomes cheap enough.
The main reason I see for having datacentres in orbit would be if power in orbit becomes a lot cheaper than power on earth. Cheap enough to make up for the more expensive cooling and cheap enough to make up for the launch costs.
Otherwise, manufacturing in orbit might make sense for certain products. I heard there's some optical fibres with superior properties that you can only make in near zero g.
I don't see a sane way to beam power from space to earth directly.
It ignores the reality of humans having memetic emotions, habits, affinities, differentiated use cases & social signaling needs, and the desire to always want to do more...constantly adding more layers of abstraction in fractal ways that evolve into bigger or more niche things.
5 years ago humans didn't know a desire for gaming GPUs would turn into AI. Now it's the fastest growing market.
Ask yourself: how did Google Search continue to make money after Bing's search results started benchmarking just as good?
Or: how did Apple continue to make money after Android opened up the market to commoditize mobile computing?
Etc. Etc.
but one of the core ideas of marx's conception of history is that human needs, wants, and human nature itself are constantly in a state of change and that those needs and desires are in large part a product of the environment in which you live, and further that humans and human society in turn change their own environments which in turn change human nature itself
Interesting, though, that per the very same article someone like Adam Smith concurred empirically with Marx's observation on the titular tendency of rates of profit to fall. This suggests to me it likely had some meat to it.
Basically dude thought the competition inherent in capitalism would cause all profit to be competed to zero leading to an eventual 'crisis' and collapse of the capitalist means of production.
Implicit in this assumption is the idea that the things humans need and want changes/evolves in a predictable way, and not in a chaotic/fractal/reflexive way (which is what actually happens).
An eventual static basket of desired goods would be the only mechanism by which competition ever could compete profits to zero. If the basket is dynamic/reflexive/evolving, there's constantly new gaps opening between human desires and market offerings to arbitrage for profit. You can just look at the average profit margins of S&P500 companies over time to see they are not falling.
The further we get from subsistence worries (Adam Smith's invisible hand has pulled virtually the entire globe out of living in the dirt), the more divergent and higher abstraction these wants and needs become, and hence the profit opportunities are only increasing -- which is how the economy grows (no, it's not a fixed pie, another Marxian fallacy).
hopefully they won't
and their titanic off-balance sheet investments will bankrupt them as they won't be able to produce any revenue