I think one thing the market may be getting wrong about OpenAI is assuming that because the moat isn’t obvious today, there won’t be one tomorrow.
$V (Visa), $MA (Mastercard), $FICO and plenty of other great businesses didn’t start with the moats we associate with them today. Visa and Mastercard began in a world where banks could, and did, create competing payment networks. FICO was essentially selling a statistical way of ranking credit risk, something competitors could build alternatives to. The product alone didn’t explain the competitive advantage these businesses would eventually develop. What made them exceptional businesses was what accumulated around the product as they grew.
The moat developed over time.
For Visa and Mastercard, it became the network. For FICO, it became embeddedness and industry standardization. For luxury, it’s brand. In other industries, scale and capital intensity eventually become the barrier.
OpenAI could follow a similar path.
Right now, LLMs look increasingly commoditized. Gemini, Claude, Kimi, Qwen, Grok, and others are very good. Switching between them costs almost nothing. On that basis, looking at OpenAI’s enormous commitments to $ORCL and asking how they will ever pay for them is completely reasonable.
But I wouldn’t extrapolate today’s competitive environment five years forward. (The Most Consistent Human Bias in Markets btw)
ChatGPT already has something extremely difficult to manufacture from scratch: massive distribution and one of the strongest consumer brands in AI.
From here, the moat could develop in several directions at once.
Users accumulate history, memory and workflows. Businesses build internal processes around the platform. Developers build on top of it. Agents make switching more inconvenient. And at the frontier, the amount of capital required to train and serve the best models could itself reduce the number of serious competitors.
None of these is a particularly strong moat individually today.
Together, five years from now, they might be.
And that’s what I believe matters for $ORCL.
Even if OpenAI eventually loses its lead, I don’t think that automatically leaves $ORCL holding the bag. The compute Oracle is building isn’t useful only to OpenAI or tied to one specific model. If the winners in AI look very different five years from now, that infrastructure can still serve whoever those winners are. OpenAI may be the customer driving a lot of this buildout today, but ultimately Oracle is building capacity for AI demand more broadly. That gives me some comfort on the downside - the bet is not entirely dependent on OpenAI being the eventual winner.
The bear case tends to take OpenAI’s current economics and compare them against its future infrastructure obligations. Of course that looks terrifying.
The bull case is that we’re judging OpenAI’s future obligations against a business that is still very early in its growth.
If OpenAI becomes one of the primary interfaces through which hundreds of millions of consumers and businesses interact with AI via inference, the revenue base in 2030 could look nothing like the revenue base we’re looking at today.
I’m not arguing that this outcome is guaranteed. Far from it.
I’m arguing that - ‘OpenAI has no moat today, therefore it will never generate enough cash to pay Oracle’ is a much weaker argument than it initially appears.
Sometimes the product comes first.
The moat gets built later. But it’s rare.
- Fermion
I love how TSA rules are different at every airport and they look at you like you're a fucking idiot when you don't speak their regional dialect of security theater.
..setzte ich mich an Opencode und mache was an meiner Website oder frage llms zu immer den gleichen Themen was. Dann will ich eine große Torte essen, aber das sollte ich nicht. Glaube eine Freundin wäre auch gut. Erstmal Klausur bestehen und Ende des Monats umziehen. Stecken..
47 Followers 4K FollowingAsymmetric research. Not financial advice. Get my real-time trade alerts and toolkits: https://t.co/AweARmByup
May trade/hold companies discussed.
172 Followers 2K FollowingFounder of @qualtrim and host of The Joseph Carlson Show • 700,000 subscribers on YouTube • Click the link below and subscribe.
5 Followers 925 FollowingFull time investor mainly US & No listed shipping + CA Mining. Use Twitter to gather info, tweet info/views that concerns my own investments-Never advice. Nomad
1K Followers 923 Followingoptions trader, here to share my knowledge and show the ups and the downs of a full time trader. NO PAID SERVICES EVER! Nothing posted here is financial advice.
643 Followers 7K FollowingFounder of @qualtrim and host of The Joseph Carlson Show • 700,000 subscribers on YouTube • Click the link below and subscribe.
4K Followers 71 FollowingHelping founders build hard hardware at Shenzhen speed. Running Powerhouse and launching an accelerator from an old factory. Physical AI / robotics.
6K Followers 3K FollowingSharing jobs at endowments, foundations, pensions, SWFs, family offices, healthcare orgs, and OCIOs. Sign up here for job alerts: https://t.co/2q94Cng28z
25K Followers 539 FollowingGoose, James Goose. vi/vim
Membering technical stuff at @openai
Made fff, @fframes_rust, and other open source software you might already be using.
21K Followers 334 FollowingHedge fund mgr. Not financial advice. Happy to debate ideas, especially shorts. River Road Partners is a made-up name for X, by the way, so don’t bother.
245K Followers 4K FollowingEverything AI + Robotics, deep dives, hottakes. I help business and individuals overcome in the age of AI. Optimist
[email protected] for partnerships 🤝
1K Followers 358 Following3rd generation Freight Broker
Vibe coding the future of logistics
Dad | Jesus | Recovered Addict 12 years clean | @freightscoutai