Steven Philips on Premarket: Apple was red, turned green - 'Ahhh, most of the world seems to hallucinate now. So why not SIRI, too? If would make it feel more human! 🙂'
on Premarket: Apple was red, turned green - 'Robert said: “serious progress being made and announced on AI-integration across Apple’s operating systems and an updated timeline and roadmap for the long-promised major and conspicuous upgrade to Siri. This includes performance on voice queries through Apple’s services platforms such Apple Music and Apple Maps.” Heartily agree, my criteria are these: 1. Accuracy – get the answer or reply as close to correct as possible. Even include a way for user to rate or vote the accuracy of reply to a prompt or request, providing QC (assuming users are honest) or AI internalized feedback on prompt request, answer, and then user follow up & acting in reply information – i.e. asking for destination and choosing answer and getting directions, then driving there. 2. Appropriate context of additional information – providing band info, history, albums, tour dates, similar music to explore, of course pricing options; map data, alt routes, additional information around route or destination, etc., all controlled by voice prompts and audio reply plus display. 3. Accuracy – no substitute for this, get it as right as possible, 95%+ and getting better with use, personalization, and growth thru feedback. Be better than MLB ABS. 4. Relative speed – if Edge, On Device Apple Intelligence takes a few seconds (Star Trek computer – “Working!”), that’s fine with me. AI datacenter communications have cellular or wifi lag plus processing time for thinking and replying, so if Apple Siri / AI can be on par then that’s acceptable performance, especially if there is NO cellular or WiFi connection and local LLM. 5. Have local Siri ask clear conversational questions on clarifying prompts if needed. 6. Have some topics off limits – Apple needs to have clear boundaries for potentially harmful or dangerous queries, offer help and guidance for better resolutions. This area would be sticky but someone has to have a compass to guide from. 7. Always, always be respectful but clear, conversing with an AI assistant, not a real human. No fantasy indulgences where users can entrap themselves. Guardrails must exist. If Apple gets flak for that, fine, because Siri and Apple should not be exposed to blame, liability, and risk of going askew. 8. Privacy – all data stays local, user centered, user managed. If the user wants to share with Apple or others, maybe ok, but first, private, especially if using user data. 9. Consistent and easily used user interface – voice, photo, display, text, document, etc. with clear navigation as needed. Ability to record or retain results as needed. 10. Relatively efficient CPU & Neural Engine overhead so as not to cause lag, hiccups, throttling, etc. except when battery is low, iPhone or device should signal user to plug in, replace battery or otherwise address power needs. 11. Accuracy plus precision in replies and data. Iterative improvements with use and feedback. Did I say that was a priority? Get it right and be best, not first, not fastest, not biggest, not largest, not whatever, be the most accurate and useful, don’t hallucinate, and user use will come, will be preferred, will be relied on, will grow with word of mouth and results. Trust is earned, not given, and Siri 2.0 has to earn its stripes to be great. Siri’s greatest obstacle is the very high expectation of what it should, could, and will do.'
on Evercore: How hyperscaler demand is disrupting Apple's precurement advantage - 'Well take this next comment as just one crazy (?) story to consider re TSMC. A Taiwanese friend says that it’s been long understood – in Taiwan at least – that TSMC will not off-shore their most advanced fabs. They supposedly do this to ensure that the US and other Asian allies like Japan and South Korea will come to their aid in the event of a Chinese invasion. With the collective West dependent on TSMC’s 3nm fabs for their most advanced chips, this would make perfect sense to me as a strategy. Ergo, and if true, then a 4nm fab in Arizona likely won’t get re-tooled as a 3. But putting in a 4nm fab in the US would likely placate any demands from Trump, at least for the time being, or pay hefty tariffs!'
on Evercore: How hyperscaler demand is disrupting Apple's precurement advantage - 'That IS a really interesting thought, Richard. One thing that quickly comes to mind why I think this might be possible is that Apple last year pledged 600 billion toward the US (mfg, jobs, etc). Of that amount the only confirmed commitments/investments we’ve heard of are those in Corning, data centers, server factory in Texas, and a few more but all relatively small in dollar amounts. Even if all those add up to 100 billion, 500 billion is a monstrous amount to spend in three years (considering Apple keeps up its end of the deal with Trump). So a huge investment toward TSMC’s Arizona plant seems logical and also sensible knowing the potential risk of a Taiwan takeover.'
on Apple shares break $290 for a new record high - 'Well, we know you can’t please the ANALyst community. Apple could deliver a cure for the common cold, and they’d complain about Apple’s negative impact on heath care economics. But Apple’s credibility is on the line this time, I think. And that’s something that’s hard to quantify, but has so far been of GREAT value to Apple. One reason for Apple’s near death before the Second Coming of Steve was they lost credibility with the user community they could actually deliver. Steve focused Apple on his ‘4 products’ and then made sure those 4 actually delivered. The best comparison that comes to my mind is Microsoft after the release of the iPhone. Ballmer & Co had no real response. If you look at Android today, it has the market position for cell phones that Microsoft has for PCs, as the primary software vendor for all the OEM phone makers. If AI turns out to be anywhere near as influential/all-encompassing as predicted, perceived weakness in AI might be sufficient to cause otherwise devoted Apple customers to look elsewhere, if they don’t think Apple can deliver the goods…'
on Apple shares break $290 for a new record high - '“There’s a lot riding on this upcoming WWDC. Apple can’t afford to disappoint.” I take the viewpoint that Apple will always disappoint someone who thinks that this is an existential point in Apple’s current trajectory. Is it important that Apple gives us optimism they are on a good path. Yes it is but given Apple’s approach to this technology it is also likely someone will say they have blown it and others will say they have played the groundwork for total domination. We are long past the time where Apple had to bet the company to introduce a product which could make or break the company. The rhetoric however has not changed as much as the company. I could joke that if anybody can afford to disappoint it is Apple. At least they have the money. Too many people seem to think they can set up a situation where Apple has to do what the writer wants or it will be a disaster for the company. My take is that Apple is setting up a situation where they are going to allow AI features to be flexible as to which AI model get’s to use them. So if you think Gemini AI isn’t your favorite you will be able to use someone else’s. Either way you will stay in the Apple ecosystem and internationally Apple will be able to deal with local restrictions more easily. Something that really does cause the company Agita given how many places have their own rules and regulations.'
on Apple stopped buying its own shares for nearly two months last quarter - 'Really great analysis, Bart! As always, the only entity that can confidently predict anything about Apple is Apple. Yes, Gurman has his spies, but I’m sure Apple would come down like a ton of bricks if they made it too obvious who they were. The very interesting possibility is, of course, that a hold was apparently put on buybacks, for whatever reason. And from the POV of long term investors, the simple ability for Apple to tailor buybacks can now be seen as having yet another enormous advantage over dividends: They can be easily dialed back as necessary whenever a more advantageous use of cash flow appears. Yet again, advantage Apple and Apple longs!'
on Apple shares break $290 for a new record high - 'Sensible observation, David! And if I mix that with the sparkling exchange Dan Scropos had with Robert Stack in the “Siri settlement” thread, I see a historic moment setting up for WWDC. Wrong or late, trillions or passion, insightful iteration or forced disruption, or one more thing?'
on Report: Trump will bring Apple's Tim Cook to China next week - 'Anything that reduces the risk of China going after Taiwan'
on In Siri settlement, Apple could pay iPhone owners up to $95 per device - 'Fair point, but at the time Jobs didn’t have the benefit of the trillion dollars that Tim Cook did. Apple seemingly missed “where the puck was going” for 15 years. No new growing hardware lines at the moment. iPhone and its associated Services are carrying Apple. Much of that goes to Jobs, while much of that *also* goes to Cool for keeping the line relevant.'
on Counterpoint: Apple's iPhone 17 sweeps the March quarter - 'Unless there is a better use for the same amount of resources. I don’t know what Apple is going to spring on us over the next 13 months, but I’m confident it won’t be a foldable.'
on Evercore: How hyperscaler demand is disrupting Apple's precurement advantage - 'I had an interesting thought and did some online reading that some may be able to shoot down 😉 Could TSMC be accelerating production of 3 nm chips for Apple by converting the existing 4 nm fab plant in Arizona to 3 nm? Apple has said it would buy 100 million chips from the TSMC plant in Arizona, which supposedly only uses 4 nm tech. The new 3 nm fab plant won’t be ready until next year. But 4 nm needs have dropped with the shift to 3 nm. However, it is possible to shift a 4 nm plant to a 3 nm plant, something TSMC is already reportedly doing. Could TSMC be converting the already built 4 nm fab in Arizona to a 3 nm one? Maybe with help from Apple? In order to get needed chips faster? And could the silicon produced during the ramp-up to mass production actually produce enough chips for Mac Neos, Minis, and Studios, which are in short supply? Possible or just a goofy thought? (It hit me about 1 a.m. 😉'
on Counterpoint: Apple's iPhone 17 sweeps the March quarter - 'Gregg wrote: “Why would Apple want to compete with a handset that doesn’t make it into the originator’s Top 5, or the industry’s overall Top 10?” I’m a longtime foldable skeptic, but if these are good and priced right, I’m persuaded they’re going to fly off the shelves in Asia.'
on Premarket: Apple was red, turned green - 'In my view there’s a lot of headroom between Apple’s new all-time highs today and several of the recently revised price targets for the shares. I believe much of that gap can be filled with Apple’s current pace of hardware innovation and serious progress being made and announced on AI-integration across Apple’s operating systems and an updated timeline and roadmap for the long-promised major and conspicuous upgrade to Siri. This includes performance on voice queries through Apple’s services platforms such Apple Music and Apple Maps.'
on Evercore: How hyperscaler demand is disrupting Apple's precurement advantage - 'PED, you have typo to fix in the title. 😉'
