I'm not familiar with Ed Zitron but failing to call the top of a bubble doesn't mean you're wrong about it being a bubble. People who were calling out the housing bubble in the 2000s were "wrong" right up until they were right. e.g. from 2006 https://www.nytimes.com/2006/01/02/opinion/no-bubble-trouble...
My own feeling is that it is a bubble: AI models are the new virtual machines. They will become commodified and low-margin hosting providers will dominate the market. Investors in OpenAI/Anthropic will lose their shirts.
Deepseek v4 flash is priced at 1/10 that of openai/anthropic. I can see a race to the bottom - or perhaps an android vs iphone split - where, the premium market is served by openai/anthropic and there is a long-tail of commodity vendors.
It's priced at 1/10, but deepseek is probably not profitable, also it's slow.
Even more interesting is the question if we would have a deepseek model without the US frontier models.
And then what's the value of the advantage that the frontier models have. It's definitely 100x more valuable to find zero days 3months earlier. Probably not in every domain but in enough domains having the smartest model is valuable.
iPhone is a consumer brand, and to some extent a fashion/status signalling choice. The market pressures in the B2B space are quite different, I expect lots of cheap good-enough models (Deepseek and others) will end up powering customer service chatbots and the like.
Who will pay 500x the price for a 1% better model? Quants and traders?
Much of the agentic intelligence is at the client. The llm backend is largely swappable. For instance, claude-code paired with any model performs well enough for many usecases. In fact, the real breakthrough is how an agent paired with an unreliable llm could perform well. Given this dynamic, I see llm tokens as the electrons or electricity, and agents as the toasters, and appliances using those electrons. If you extend this analogy, value will bubble up into the appliances which would each have consumer preferences. A token is a token no matter who produces it, just as an electron is, but I like my KitchenAid toaster, whats your preference ?
But Claude Code works about as well as Codex and roughly the same as Github Copilot. So what part of this pipeline is supposed to command premium pricing?
You might prefer your KitchenAid toaster but I'd wager you won't pay enough to support a trillion dollar valuation.
> They will become commodified and low-margin hosting providers will dominate the market.
I have my doubts about this. We have not seen a viable YouTube alternative because the underlying costs of handling video content are significant and YT has custom hardware and sophisticated software. When we look to the broader cloud market, hyperscalers dominate. We are likely seeing similar when it comes to Google's TPU and access to Nvidia's best offerings.
That being said, I did just pick up a DGX Spark and it runs qwen-3.6 sufficiently well to be a viable interactive coding assistant. Certainly more than enough for unattended agents.
YouTubes biggest moat the last 10 years is probably more that all the viewers and creators are already there. Any competitor has a huge disadvantage - creators are not interested in a place without viewers, and viewers not in a place without creators/content.
yeah, network effect is real, and you cannot get viewers without competitive video delivery, so perhaps the moat is more like having an ocean on both sides
Network effect means it will be a huge and risky undertaking, and one needs to solve the bootstrap problem. But the costs of video delivery means that one would have to burn serious cash in the meantime. So it works in tandem.
TikTok kinda did manage to make a dent though - I suspect it substitutes for YouTube in some cases (though not all).
arguing that the reason youtube is succeeding is because of video hosting costs is hilariously misguided.
the content and creators are the only competitive advantage they have. there are MANY video hosting platforms out there but they just don't have the content to attract large audiences like youtube does. they have a strong early mover advantage
Definitely, Google Vertex Ai serves up other companies models better than they can themselves. The TPU is the bee's knees. I really hope Google makes a take on the DGX Spark
Discoverability of other content and ad money. And then critical mass of viewers leading to sponsorships and other exploitative models of monetising outside Google.
Ads might be questionable model for lot of use cases. And network model only works for promotion but does not lock users in because content is only available in one place.
It is unlikely that models will have network effect because (1) there is less of a two-sided marketplace and (2) people are already forming brand preferences. We also see significant convergence among the agent harnesses as well.
I'm currently building out an internal agentic orchestration platform for business and development and a requirement is to support multiple models and tools so people have an amount of choice.
- If costs are going up: oh no AI is not sustainable! Bubble bursting!
- If AI costs are going down: oh no it’s getting commoditised, OpenAI bankrupt anytime soon TM
- If companies have moat and get bigger: oh no companies are getting powerful! It’s bad and we must oppose them because they can rug pull anytime and enshittify!!
What situation is something that you would be okay with? Because people seem to have a problem with any outcome.
Huh? What makes you think I'm objecting to anything?
I'm actually pretty happy if we have a competitive market for AI that maximizes consumer surplus. For a while there it looked like AI might remain in the hands of two or three corporate giants.
Yea no worries, just wanted others to know since when I typed the comment, the submission was quite fresh. It was my hope that it would make less people confused.
Actually, it's deterministic -- our product doesn't move money, so when the user gives us access through Plaid, we're only getting read-level permissions. We actually don't even get full account numbers.
A company working with Plaid has to request separate product "scopes" through Plaid in order to be able to move money.
I'm not that familiar with Plaid, but if it works like Yodlee, users have to hand over their credentials so there's no real security, it's just that their scraper is designed to be look not touch.
Plaid has OAuth-based access for most of the big institutions now, but yes, for smaller institutions, they do scraping. Thankfully, Plaid's been around for a while now and has a good track record. It would be a non-starter to give your credentials to a small startup directly.
The awesome thing is, if it's something you're interested in, you can basically set this up yourself -- a local AI model, some browser automation, and it does mostly work! It's just a pain to keep running. But it's definitely doable.
I am the kind of geek who doesn't even want a second record of their finances around. My physical bank statements even have a procedure for how long they live until being shredded.
If I were interested in this though, yea, a local LLM would be the only way.
doesn’t change the bigger, more important fact that the struggles people have with git stem from the system design. i.e. the thing that ultimately determines what commands people need to run in what order (see points 2 and 3).
Arguably yes. git has a terrible developer experience and we've only gotten to this point where everyone embraces it through Stockholm syndrome. If someone had been looking at analytics from git, they'd have seen millions of confused people trying to find the right incantation in a forest of confusing poorly named flags.
reply