[liamconnell]

Unfortunately for OpenAI, they are not positioned to capture value from any of the "big margin" use cases that they highlight as key to their future. I think all of these are pretty unrealistic for them:

- Revenue sharing from drug discovery (called out by OpenAI CFO): Why would a pharma company give away the upside to a commoditized intelligence layer? Why would OpenAI have a more compelling story than Google Deep Mind, which has serious accolades in this space?

- Media generation for ads and other content: For ads, OpenAI is facing off against Google, Meta and Amazon, all of which have existing relationships with advertisers. For the foreseeable future, AI content will be a major discount product compared to humans. OpenAI will not get to charge $1M for an ad like a production company does. So the TAM of ad production (~$50B) shrinks below $1B because AI deflates prices so much.

- Other agent use cases: OpenAI doesnt have a surface to build these on. Google has chrome, Microsoft has office, Apple has OS's. The other use cases like coding will be a low-margin competition between model providers until some of them throw in the towel. The players with the best cash position win - and thats not OAI.

I think the place that they could win is retail (also called out by OAI CFO). They made deals with Etsy and other small retailers. I was fixing my guitar the other day and would have instantly bought the tools it had suggested that I would need. The problem is that they have to win against Amazon here, and there is zero chance of a partnership for obvious reasons.

[vb-8448]

At some point, I'd expect the following pattern to emerge:

- random user: hey chatgpt, I need a new mechanical keyboard, buy me one - openai will get money for mechanical keyboard vendors to be on top of gpt's agent list

the ad business will shift from trying to hack google to hack gpt

[liamconnell]

Yes that's what I meant. But Amazon will fight for its life to stop this. So OpenAI will have to go to other retailers, which don't have the same product catalogue size as Amazon. Best case scenario: Target or Walmart. But there is a reason that OpenAI announced deals with Etsy and Shopify rather than those.

And OpenAI doesn't have as much product insight as the retailers so they have to rely on the retailer to choose which is the "best" mechanical keyboard for this person. And at that point, pretty much all of the shopping value is being provided by the retailer rather than ChatGPT, so why would they get much money?

There's a market for this but its not going to be trivial for OpenAI to win it. And it probably wont be a cashflow monster like AdWords or Amazon.

[rco8786]

> Amazon will fight for its life to stop this.

Why? Amazon advertises heavily on Google search, why wouldn't they do the same with OAI?

[liamconnell]

Good point!

On the other hand, Historically Amazon didnt compete with Google (until GCP). They do compete with Microsoft, which is pretty closely aligned with OAI. They also have large investments in Anthropic.

Even if OpenAI did win here, would it be a profit monster like Google Adwords? Adwords had the auction model which meant that certain categories were hugely lucrative for Google. Can a chatbot do the same? If I know that the product I buy is simply auctioned off to the highest bidder, what's the point of using an agent to help me shop? There has to be a pretext of the agent actually looking out for my best interest, otherwise I would just use search. Nobody expects adwords to look out for their best interest. They are always free to skip the ads section if they choose.

It will be hard for ChatGPT to implement an auction model since it will be different for each product category. Hiring a lawyer will probably have a different interaction from buying groceries. On Google+AdWords, its all just search results and ads.

If there is no auction, then all of this is WAY less profitable than the Google model. So once again - not going to save OAI from negative margins.

[rco8786]

> Can a chatbot do the same?

I really don't see why not. If anything, there's even more user interactions with a chatbot which means more opportunities and more context for placing targeted ads.

And why wouldn't they do an auction? There's nothing stopping them. In fact, again, it's even easier because users are not conditioned to near instantaneous results from a chatbot like they are from an internet search. Internet search also has different categories, I'm not seeing why it's a particularly different or more challenging problem re: hiring a lawyer vs searching for groceries.

[Analemma_]

Because they make a shitload of money by arranging Amazon search results in a certain way, selling favorable placement on those results, and inserting upsells ("Try Amazon Prime!") into the checkout process, all of which are at risk if Chat-GPT becomes a frontend to buying on Amazon. (Note that they are already vigorously going after Perplexity for trying to be an Amazon frontend)

In general, Big Tech will never allow itself to be just the backend to a service where another company controls the frontend and the relationship to the customer. That's how you get commoditized and ultimately replaced.

Examples: you cannot get a streaming box with universal search ("which streaming service has show X? Just hit play and go"): the streaming services staunchly refuse to provide the APIs to do so. Nor is there interoperability across messaging apps to let users supply their own frontend clients. AI and MCP will go much the same way, it will be locked down as soon as it presents a business model threat.

[glpgeFwac]

Agree on your overall point, minor note that Apple TV does decent at being a streaming box with universal search. The benefit of buying into a walled garden is that sometimes platform owner and user interests align.

[rco8786]

It's not just Apple. Every major platform has universal search now.

[rco8786]

> all of which are at risk if Chat-GPT becomes a frontend to buying on Amazon.

Can you explain to me how this is different than the literal world we are currently in, where Google and web search serves as a frontend to buying $billions of products on Amazon already?

> you cannot get a streaming box with universal search ("which streaming service has show X? Just hit play and go")

You've clearly not used a Roku TV, Apple TV, or Amazon Fire Stick lately :). Universal search with "click and play" and deep links into individual streaming services is table stakes for streaming TV now. Your statement here could not be more wrong, honestly.

[aeturnum]

Because amazon doesn't have a web search service but they do have a product recommendation service? Even if they do pay OpenAI they would certainly be competing with their own service and keeping prices down via that. OpenAI needs Amazon (or some other fulfillment company) to deliver products. Amazon does not need OpenAI - they can build their own recommendation engine or work with another.

[rco8786]

> Because amazon doesn't have a web search service but they do have a product recommendation service?

That's the whole point here. People use web search as a product recommendation service, even though Amazon has one natively. What makes you think people won't (and they already are, in massive numbers) use chatbots for product recommendations and web search?

[chickenegg]

But OpenAI has the attention. It's where people ask for product recommendations, and it has context about the user. Surely Amazon doesn't need OpenAI, but OpenAI will be another valuable distribution channel for them - unless some other LLM takes the crown.

[jsnell]

> Best case scenario: Target or Walmart. But there is a reason that OpenAI announced deals with Etsy and Shopify rather than those.

I'm confused at that statement. Walmart and Target are exactly who OpenAI announced deals with recently:

https://corporate.walmart.com/news/2025/10/14/walmart-partne...

https://openai.com/index/target-partnership/

[liamconnell]

Didn’t see that yet, interesting!

[martinald]

I'm not sure. Amazon isn't usually hugely price competitive nor does it have stuff I can't find elsewhere.

What it does have is very high convenience (I'm signed in already, and I know the checkout process by muscle memory). To be fair it also has excellent customer support, but I'm not sure I would go out of my way just for that (I return a handful of purchases a year out of 100+).

These go away with 'agentic commerce', at least in theory, because the agent/MCP/API does this for the user.

The other advantage it has is excellent logistics, but that's more of a benefit for Amazon than the user IMO. Lots of small ecommerce sites can have 'excellent' logistics, because they are much smaller. The only unique thing Amazon has in the UK at least is same day delivery, but I believe they lose a fortune on that and really try and push you away from it. This may vary where you are but in general next day delivery works great in the UK from most sites (DPD/RM Tracked 24). Gets a bit hairy with 'economy' delivery from Evri or Yodel tho.

[luckylion]

> But Amazon will fight for its life to stop this

have you seen Amazon's "Rufus"? It's hilariously useless.

[vmurthy]

> have you seen Amazon's "Rufus"? It's hilariously useless.

I'd argue -- for now. Maybe it's an incentive/urgency thing. At the moment, Amazon isn't seeing ChatGPT do the buying of goods bypassing Amazon's own search. I expect Rufus to drastically improve especially given that Amazon has an AWS offering of LLM(s) [0].

[0]https://aws.amazon.com/ai/generative-ai/

[notatoad]

this was supposed to be the business model for alexa. amazon had all the pieces - they had the product listings, marketplace, ordering infrastructure, the smart speaker in your home always listening to you, and they built exactly that product.

it hasn't exactly taken off, and i don't think OpenAI has addressed any of the problems that prevented amazon's version from being a success. and that was without taking advertiser money to choose which product to sell you, amazon was happy to just make a sale. if the product choices the AI shopping assistant makes are driven by advertiser dollars instead of product quality, i really don't expect consumers to accept it.

[danpalmer]

What do you see the problems that prevented Amazon as being?

I don't know much about this, but I'd have thought it was the lack of display or ability to critique the choices Alexa makes. But ChatGPT doesn't have that problem because you can see and "discuss" the buying decisions.

[array_key_first]

As soon as the cat is out of the bag and consumers know that ChatGPT is a vehicle for advertisement, consumers will reject it.

If I, a consumer, want to buy a car, I need to do research. Where do I go? Online, across many websites. I talk to my friends. I talk to my coworkers.

Where do I NOT go? To the car salesman, and ask him for help. Because of course he will lie - he's a car salesman, he wants to sell cars that he sells.

Even with Google we see this being the case. Nobody is clicking the Google ads at the top because they know those are ads, not research. They only do it accidentally, which is evidenced by Google making it more difficult over time to tell what is or is not an ad.

[chickenegg]

> Nobody is clicking the Google ads at the top because they know those are ads, not research. They only do it accidentally

I think you're evaluation of how many people click on Google ads and for what reasons is quite off. I'm sure you and most of the people in your circle are like that, but that's not how the vast majority of internet users behave. Google isn't generating $200 billion annually of accidental clicks.

[array_key_first]

Yes, yes they are generating $200 billion on accidental clicks.

Consumers don't know what they're clicking on is an ad, that's the only reason they clicked on it. They googled something for an answer - and oh look, the top results are what I want! Except those aren't results, they're ads. Here, put on your reading glasses - there's 8pt font there that says it's an ad.

It's the same thing with YouTube. Maybe half of the ads are just outright scams, and consumers know that. It's not like TV. So they don't click them, because you're playing Russian roulette if you do that. They only click them accidentally.

The main problem with online ads is that Google and Meta are dumb as rocks. They've decided to sell ad space to anyone with a pulse for a quick buck. The problem with that is that, over time, it devalues the ad space as consumers are trained to ignore it. We're quickly approaching the point where these ads only work on people born yesterday.

The ad space is grossly, grossly overvalued. Shh, nobody tell advertisers teehee!

[mlrtime]

I'm with you on a Paid model, but if the future's best model was available for free, unlimited use, but you just agree to a few brand placements.

I could see a majority of non tech (non HN) people take that deal.

[giantrobot]

> But ChatGPT doesn't have that problem because you can see and "discuss" the buying decisions.

If OpenAI is accepting money from advertisers to push products then ChatGPT is just a salesman. You won't be having "discussions" you'll be actively sold stuff at all times. What an awful yet banal dystopia.

[Duralias]

Never used Alexa to buy something (not even sure that was supported here) but it not showing you what you are buying, which I feel like it must have on some devices, just sounds like a design mistake, ChatGPT and all other LLMs will be the same if just spoken to.

Also, I would never discuss something I am buying with an LLM, the moment advertising starts being used to influence its output it will be the same as discussing the product with the product page (which of course is only positive) and ignoring negative reviews.

[jimkleiber]

It's already shifting. I know someone who used to do SEO and is now marketing how to get in llm results.

[Analemma_]

This was Amazon's huge bet on Alexa: that if you made a frictionless way to buy products by saying "computer, buy me [thing]", then people would use it and then you could sell favored placement on it.

It was a total failure. I know lots of people-- both technical and non-technical-- who have Alexa devices, and not one of them has ever bought anything with it. You can read various comments from Amazon insiders confirming that the rate of buying things with Alexa was close to zero. And why not? It's the shittiest possible way to shop, like buying a lottery ticket except where the RNG is knowingly gamed. This is why Amazon is writing off Alexa entirely.

I've commented to this effect before, but "what if people could shop sight unseen" is a PM fantasy, not a thing anybody actually wants. LLMs might be useful for helping with research and comparison shopping, but the "one-click [or one-prompt] buying" workflow is not gonna happen.

[onesociety2022]

Favored placement is never going to work but I would use Alexa for repeat purchases: “hey Alexa I bought some Crocs shoes a year ago. Can you reorder it?”.

Or purchases where I know exactly what I want but don’t want to search and add to cart manually: “buy a new 3 foot USB-C braided cable from Anker”.

[addaon]

> “hey Alexa I bought some Crocs shoes a year ago. Can you reorder it?”

But this is exactly what Amazon can't do. Basically all Crocs (crocss, croks, crox) on Amazon are counterfeit, and they don't even have a record of which ones they pulled out of the bin last year to send to you so they can try to grab the same countefeits; and the company that listed them a year ago is probably on their fifty-second name change since then, and the "Satan's anus green" that you chose because it was half the price of the other colors is now "Satin Annux Green" at 2x the price of the other colors…

[onesociety2022]

I have been ordering the same Crocs from Amazon for about 8 years now. I don’t think I have ever received counterfeit Crocs from Amazon. I basically go to “my orders” page in the app, search for the last order and then reorder. I do have to make sure they have not jacked up the price because the price fluctuates quite a bit for that shoe I buy. And I make sure the ratings for the third-party seller are very good before ordering. They can definitely have Alexa automate this entire workflow (“hey the shoes cost $20 more than last time. Do you still want them?” Or “the shoe is only available from a different third-party seller. Do you still want them? This third-party seller only has a 90% positive rating”).

[addaon]

How do you find out if the third-party seller you're choosing is correlated with the product that they choose to deliver, or if they're co-mingling inventory? You used to be able to get an answer from support in about 30 minutes or so, so for several years I spent the 30 minutes per order to try to find out, but recently they've claimed they can't say.

[kristo]

At that point why would I use ChatGPT if it’s just getting paid to show me stuff?

[stevenwliao]

Same as any other ad supported services- the service is useful or entertaining.

[____tom____]

and then, to make money, they'd have to stop giving the correct answer, just like search engines

[rco8786]

This is it.

Their real moonshot should be search and ads. They're already taking big chunks from Google, they're just not monetizing at all (yet).

I already use chatgpt constantly for product research.

[svara]

Yes but this is incredibly competitive and undifferentiated.

It's a huge market but who will it be a profitable business for?

Likely a company or multiple who own some sort of platform that people are already on, so not OpenAI.

What they have right now is the strong ChatGPT brand and that does mean a lot. But how long will it last?

They're not the technology leader anymore, and that spells a lot of trouble.

They are at a stage where they need to dominate the market and then leverage the data that gives them, plus the brand, plus the tech advantage to establish a durable near monopoly, but it looks like it's not working.

It's a bit as if in 1999 3 equally strong Google competitors had popped up, with some pulling ahead.

[rco8786]

> It's a bit as if in 1999 3 equally strong Google competitors had popped up, with some pulling ahead.

You mean like Yahoo, Lycos, Altavista, etc?

[svara]

They sucked compared to Google.

[brazukadev]

That would be the optimal scenario for openai but even this one I'm not really expecting to happen anymore. OpenAI failed.

[javier2]

I am already using claude to help me shopping. It can be so hard to find the actual specifications of a product. Amazon is filled with nonsense information, and its nearly impossible to compare different variants of things like monitors, tvs, cpus and other technical things that clearly are made with certain specifications in mind.

[dboreham]

I use it for gift suggestions. So they could easily add affiliate links on the page.

[holden_nelson]

> OpenAI is facing off against Google, Meta and Amazon, all of which have existing relationships with advertisers.

I will point out that these companies have existing relationships with advertisers because they have massive, sticky userbases and advanced targeting tools. The average consumer is absolutely using ChatGPT for personal use, and maybe Copilot at work if applicable. And they're using Google's AI by proxy when they perform searches.

If OpenAI were to roll out advertising tooling, I have no doubt advertisers would flock there to try it out.

Additionally, the other thing I think OpenAI leads in is Product. Google is amazing at creating technologies and awful at creating products. I think OpenAI can be positioned to win based off of that alone.

[autoexec]

> . The average consumer is absolutely using ChatGPT for personal use

In my experience the "average consumer" isn't doing anything with ChatGPT except maybe play with it for a little bit before getting bored. They actively avoid AI when the apps and products they use try to shove it down their throats and they search the internet and ask their tech savvy family members for ways to disable AI in their stuff when they see it nag at them about using it.

Inevitably, AI ends up being used by people in some ways (like the AI reply at the top of every google search) but almost never because the average consumer asked for that or wanted it. It's a toy when they want to use it, and annoying when they don't but are forced to.

[benhalllondon]

But they have 700million WAUs?

[hattmall]

Yeah, but the report citing that number doesn't make clear how they calculate that. But maybe that's just a failure of my comprehension though I know some people like to spin words. So is 700 million unique users each week? Or is 700 million unique users that were active in A week, but not necessarily the same week, as in a creative way of saying weekly instead of monthly? The report simply states "in September of 2025", but September isn't a week. Or is it not even unique but simply distinct users in a week. Would the same person on Monday and Tuesday be 2? Or even Monday morning and night? Or any different session. The report citing the 700 million specifically seems to leave out word unique which is pretty key to any meaningful statement on visitor usage.

They go a bit more detailed in the end to say 2.5 billion messages a day.

[holden_nelson]

Eh I definitely agree this archetype is real but I disagree that it’s the one that constitutes the average consumer. My dad is a carpenter and my mom is a nurse. My wife is a hairstylist. None are particularly tech savvy. All three use ChatGPT quite a bit. Stuff they would previously google. How do I make an apple pie? Should I see a doctor? Stuff like that.

As another commenter stated, ChatGPT has over 700 million WAU. There are only 4.4 million SWEs in the US. I think it’s caught on

[aylmao]

> Additionally, the other thing I think OpenAI leads in is Product. Google is amazing at creating technologies and awful at creating products. I think OpenAI can be positioned to win based off of that alone.

I agree that Google isn't great at creating products anymore, but I'm not sure that OpenAI is. We've seen relatively simple products by them (a chat app, a short-form video app, various web interfaces) but we haven't seen anything as complex as some of Google's bigger products (Gmail, Docs, Maps, etc).

If OpenAI hits jackpot with a "simple" product, it could be easily replicated by a bigger company in the way Meta quickly copied Stories from Snapchat or TikTok to make Reels. It's already happened with Chat; the LLM is hard to compete against but the actual product, a web/app chat interface, was quickly copied by other companies with LLMs.

OpenAI would need to make something very complex and hard to copy to give it a solid head start they could really build a moat around— something like Google Maps, which took Apple years to replicate (and other companies won't even try to) or the iPhone, which was years ahead at launch. I just don't think we've seen OpenAI prove it has the capacity to build a product like that yet.

[famouswaffles]

>If OpenAI hits jackpot with a "simple" product, it could be easily replicated by a bigger company in the way Meta quickly copied Stories from Snapchat or TikTok to make Reels. It's already happened with Chat; the LLM is hard to compete against but the actual product, a web/app chat interface, was quickly copied by other companies with LLMs.

IG reels never became as popular as Tiktok and did basically nothing nothing to peel users away from Tiktok. For a long time, it was a meme that IG reels were just copy pasted tiktok content. Similarly, Meta's LLMs are used so little they honestly don't even register, despite being stuffed into everything they own, apps with billions and billions of users. Gemini is doing well but it's still a very very distant 2nd, despite being automatically downloaded and nudged in android phones, a platform with billions of users. Microsoft is by far the biggest player in consumer laptops, with edge and bing being the default options. So why can't they come even close to chrome and google ?

Time and time again, we've been shown. You can copy all you want, you can even shove it into the faces of your billions of users and find use for it. Doesn't mean you'll beat the market leader. You'll rarely beat market leaders just by copying them.

ChatGPT is the 5th most visited site on the planet. No other Consumer LLM service is remotely close, regardless of how many billions of users the entrenched players are shoving their copies into.

[aylmao]

> Time and time again, we've been shown. You can copy all you want, you can even shove it into the faces of your billions of users and find use for it. Doesn't mean you'll beat the market leader.

Except this is not true.

Microsoft became the biggest company in the world with this strategy. Most famoulsy, Windows was a copy of MacOS, and Internet Explorer (which became so dominant it took Microsoft to court) of Netscape. Android copied iOS and became bigger than it. Again, Instagram famously copied Stories from Snapchat, and quickly took over [2]. There's many examples of a bigger company copying smaller ones and using their distribution or ecosystems to win.

> Microsoft is by far the biggest player in consumer laptops, with edge and bing being the default options. So why can't they come even close to chrome and google?

I don't know if you're perhaps too young or too new to the industry to know about the browser wars and the Microsoft anti-trust case, but if Microsoft pulled all the stops it once upon a time did, it would. It so happens Google is has an infinite money machine too though, and can pour many millions in the courts if Microsoft were to try.

> You'll rarely beat market leaders just by copying them.

Let me correct this sentence for you. You'll rarely beat the big companies by copying them, but the big companies can easily beat you by copying you. OpenAI might seem big, but it hasn't found its infinite money machine yet.

[1]: https://www.forbes.com/sites/tomward/2017/05/03/oh-snap-inst...

[chickenegg]

I know plenty of SaaS companies that are paying tens of thousands of dollars every month for LLM optimization & AI visibility. Lots of marketing agencies that have been working hard on delivering paid ads & search engine optimizing are floating in a river of cash now because all these companies are panicking. So, yeah, fully agree that if OpenAI rolls out a halfway decent advertising option, advertisers will throw money at them.

[kristianc]

Which is likely why they won't try. Trying to raise that much money from 20 large individual customers would be suicidal. At the scale they're talking about, you need billions of "customers".

Trying to embed themsleves into every enterprise workflow and taking a cut from it seems much more likely than them trying to invent the next killer app. ChatGPT is just the marketing arm which keeps them front of mind.

[jstummbillig]

> - Revenue sharing from drug discovery (called out by OpenAI CFO): Why would a pharma company give away the upside to a commoditized intelligence layer? Why would OpenAI have a more compelling story than Google Deep Mind, which has serious accolades in this space?

I am not sure I follow. They "give it away", because they have to. They have to pay any of the model companies. What do DeepMind's accolades matter if it's commoditized, as you propose?

AI resources will remain scarce for the foreseeable future: I have to literally wait multiple Minutes to get an answer for semi-hard coding problems. The current demand is the delta between this, and the few milliseconds that it could take if supply was there. I suspect the tension will grow. Why would there not be multiple companies positioned to capture value? Assuming that any of them can turn demand into profit, that seems to be the most likely story right now.

[liamconnell]

The CFO isn't talking about selling tokens to pharma companies. There's no money in that. She proposed revenue sharing. In this scenario, OpenAI's AI service helps discover drug candidates, and shares the IP ownership of the candidate (which is basically a risky bet that it will get through clinical trials and be profitable). Biotech is a complicated market filled with smart people and great negotiators - they dont give away IP ownership without a lot of thought.

If OpenAI wants anything more valuable than selling tokens, they will need to offer something valuable and differentiated. Right now they are not differentiated in the space at all. Look up "OpenAI Biotech" - anything that they've built themselves?

If any company will have a new product that biotech companies will pay top dollar for, its Google. Deepmind has been in biology (proteins) for almost a decade and they it has subsidiaries like Isomorphic Labs that are bringing products to market.

[semi-extrinsic]

> OpenAI's AI service helps discover drug candidates

This has never been the difficult/expensive part of drug development.

[mattmanser]

His point is AI's already getting commodified. So OpenAI won't get a portion of the profits or revenue sharing, it'll be a simple transaction. They simply pay for compute time.

It's like pretending sulphuric acid manufacturers would get the right to demand a portion of drug company profits.

[jabberwhookie]

Big pharma might choose to pay full cost to get reasonable speed. To get to a partnership that looks a lot like tenant farming you would need a model that is actually 100X better at drug discovery than any other model, Why would that be OpenAI instead of deepmind? Not that either is likely worth much premium.

Generally, I think only penny stock pharma cares at all to deal with IP with any kind of baggage instead of having already forgotten it in the backlog.

[danpalmer]

> Why would a pharma company give away the upside to a commoditized intelligence layer?

Why would a commoditized intelligence company give away the upside to a commoditized silicon company?

It still amazes me that Nvidia is worth so much, they're just one slice of the value chain, from mining, through chip fabrication, through to chip IP, through to technology stack, training, inference, and product integration.

I understand the reasons why, it's mostly lock in with CUDA and isn't really about unique chips, and I think the market sentiment on this is changing, but still it's crazy to me.

[ivan_gammel]

>Other agent use cases: OpenAI doesnt have a surface to build these on. Google has chrome

ChatGPT app is their Chrome. Large consumer base using chat on daily basis can expand to prosumer and to enterprise. They build an emotional connection to their customers that has the vibe of iPhone.

[liamconnell]

Yes but its not relevant to the agent use cases (which are mostly about interacting with external systems). So agents built by Microsoft (Copilot) can natively interact with Office files in Sharepoint, and the Sharepoint product team can build to enable this in special ways. OpenAI has to use the APIs and deal with rate limits, speed issues and other limitations.

[ivan_gammel]

Do Oracle, SAP or Walmart invest in their own models or they build integrations? There are lots of companies which don’t have other options but to work with 3rd party LLM vendors. OpenAI, Anthropic and Mistral are better positioned for this than others.

[hobofan]

A large portion of those customers buy those models from their preexisting cloud vendors: AWS, Azure, GCP.

The customers are also price sensitive and are for many use-cases largely fine with last-generation model performance if that means they are cheaper. With that, there is little moat for the model creators, forcing a race to the bottom for model licensing, and the biggest chunk of the profit being captured by the cloud providers.

[throwway120385]

Oracle, SAP, and Walmart already have teams of people managing the data they have in-house. I find it hard to believe that those people can't pivot to working with an in-house LLM under the right leadership.

[ivan_gammel]

Data landlords have no reason to invest in their own models when they can license their content to all vendors and offer integrations with them. Their business models are enhanced by AI, not threatened by it. For Google and Meta it’s different: they offer entry points to data and chat apps are direct competitors. So chat apps with good integrations can be as big as largest ERPs, for example.

[9cb14c1ec0]

> OpenAI doesnt have a surface to build these on

Correct. They certainly could. An OpenAI alternative to g suite and MS Office would be a good start (integrated with the chatgpt mobile and web presence), but would also be a huge engineering effort.