As long as you can download apps and set permissions for communication between apps/devices, then it seems like 3rd party developers will make Apple devices more valuable for them.
OpenAI is at the state it is in, because of massive amount of data is hoovers from the internet. Apple is culturally not set to do that kind of work.
Imagine an Apple+Open AI releasing Sora, which can produce images based on [christopher nolan] or [another-famous-visual-style], personalized the way users want.
And no, you cannot do that on-device and get away, because that is (if at all possible) is several years away. Apple has to demonstrate leadership right now, and the only way to show that is by dealing with data, a ungodly amounts of it, which Apple has themselves said to be radioactive.
Apple painted themselves to a corner, and now they are paying the price for it.
Financial facts: 2024 Apple Revenue: US$391B, income: US$94B. Apple currently has a market cap of 3 Trillion and cash on hand of $49B
From article Mentions TAM for AI of $1.8 trillion by 2030.
Pick a profitability for OpenAI: then look at whether it would make sense for OpenAI to sell to Apple, or how Apple could buy OpenAI.
Apple already gets 30% cut of revenue for Apps.
Maybe the author thinks that OpenAI should buy Windows Mobile from Microsoft?
msgodel•7mo ago
Their problem isn't inability to develop or lack of access to technology, they're extremely good at that, They're like a vertically integrated combination of Microsoft and Nvidia.
Their problem is the lack of vision and inability to productise what they develop. IE their problem is that they're Apple, buying other companies won't change that.
znpy•7mo ago
At those levels you don't buy other companies to only get the tech, you also acqui-hire the talents.
msgodel•7mo ago
They don't have to though. Language models are commodities and the weights are free.
bitpush•7mo ago
That's like saying chips are a commodity. You can buy from Intel, or AMD. Why bother making something yourself. And yet, Apple has proven that unless you vertically integrate you're not going to differentiate.
msgodel•7mo ago
Can you get a slight edge with the particular RL tuning on some models? Sure. Does it actually result in qualitative changes in what the model can do? No not really, all of them can do MCP, all of them still hallucinate etc.
znpy•7mo ago
You're probably missing the point. The output is certainly good when compared to the number of people involved. But if you need "more output" at some point you'll need more people.
Also, new people usually bring new know-how. No company ever died from having too much know-how.