HilliTech's Newsletter: the talking donut won't save you
It seems the repeating topic of Apple and OpenAI's lawsuit will continue until we reach new iPhone season, so bear with me. Each piece of news just hints hat the horrific mismanagement of one of the biggest companies of this decade.
Here are some stories I'd like to share this week:
Luke Plunkett
Wesley Hilliard
Wesley Hilliard
Someone is scrambling, and it isn't Apple
Reporting around Apple's move into AI has been quite the drama in recent years. People wanted to form a narrative, and they succeeded at that. The created narrative was one of incompetence, taking the wrong paths forward, and a mass exodus driven by a set of leadership that didn't understand AI.
Yes, 2024 was an attempt that had a setback. Reports tried to suggest that Apple ignored AI as a passing fad while somehow forgetting that Apple's internal AI/ML team was established in 2018.
Giannandrea and team landed on the on-device concept early on and targeted it for the past decade. Apple did ignore the AI industry trends at large, yes, because it didn't aim for server-side first and scaling down, it aimed for small utilities that could scale up.
Remember when everyone said Apple would suffer because it started closed and slowly opened things up while Google started open and closed things down? Similar discussions were being had about the viability of iOS, Apple's stance on privacy, and its ability to hold onto a market when more people were buying cheap Android handsets. Sound familiar?
On-device, private, secure. These concepts held true throughout, which contradicts the narrative that Apple didn't know what it was doing and leadership was "scrambling."
Sure, many employees were poached or left Apple, but their churn wasn't particularly notable or unusual for the time period. Compared to other tech companies during the AI boom of the last few years, Apple had incredible retention rates and lost very few employees compared to the total.
Interestingly, the lawsuit with OpenAI shows exactly how the narrative given to Apple was actually the one occurring within OpenAI. Desperation, a race to get some kind of monetize-able product to market, hiring Apple employees and allegedly encouraging trade secret theft because they had no plan and no ideas. And now their ultimate solution is a donut with flashing lights for $400 that can perform the same tasks as your iPhone.
When examined from the 1,000-foot view, Apple has held strong and consistent for the past six years while OpenAI's desperation has shown through at every turn (remember they were going to monetize porn bots? What about that wearable thing?). It's kind of incredible and many books will no doubt be written about the insanity in a decade.
Apple was working on AI technology and training some very basic Apple Foundation Models using the technology and skills at its disposal. It scraped the web with AppleBot, though it attempted to do so responsibly. However, instead of attempting to scale quickly and deliver a frontier model that could be split into coding agents, search agents, and autocorrect, it focused on small local models and an ML backend to hold Siri and everything together.
Executives were clear that Apple had no intention of building a chatbot. That language only changed once Apple realized the hybrid approach of tying AI and ML together wasn't going to work. That's when it canceled the contextual Siri launch.
I'm aware there are arguments that contextual Siri never existed. I'm not buying that. Apple clearly had something in development and testing, but whether or not it was ever ready to be shown to the public is another question entirely.
When Apple pivoted in 2025, it didn't scrap everything it had built before and take on a whole new approach. That would indicate it felt Gianandrea was on the wrong path. No, it doubled down and sought out a partner that could strengthen what it already built and deliver some of what customers expect from AI in 2025/2026.
OpenAI was already showing signs of being disgruntled with Apple. Anthropic was asking for too much money, and Google already had a lucrative partnership with Apple. Of course Apple went with Google. Apple had also used Google Tensor technology when developing the first version of Apple Intelligence.
The Apple Foundation Models that already existed were enhanced through distillation from a Gemini model that was built to run on Private Cloud Compute servers. The end result is a set of models that can continue to be enhanced that are wholly Apple's property and technology. It isn't interfacing with Gemini Assistant, Google Search, or any Google service.
I find it interesting to see how much a perceived narrative can be created from months of targeted reporting that skirts around facts and half-truths. The reality was always more interesting, but for whatever reason, likely related to money, people like to see Apple fail. Or, at the least, report Apple failed even when they haven't.
Google Assistant is shutting down, Alexa is worse at nearly every aspect of its original functions, and Cortana no longer exists. Siri is the only remaining smart assistant from the original era because its competitors had to seek a rebrand to avoid user's dislike or distrust of older products or to get more AI investor capital because "it's not the old thing, it's AI" marketing.
Apple's upgraded Apple Foundation Models power Apple Intelligence and Siri AI. Yes, Apple tossed AI on the end of Siri, not as a rebrand to get people to disassociate it from the original, but to let users know when they're talking about the latest version. I wonder if Siri AI will become colloquial, or if even Apple will fall back to just calling it "Siri." (Think Apple Home vs HomeKit, it's still HomeKit as a standard, but Apple Home is the household name now)
I even saw some people lament that Apple wasted Apple Intelligence on the 2024 products as if the name was going away or had a bad association. Not only is it still there, but everything Apple announced in 2024 is enhanced, plus, the contextual systems based around app intents shipped too.
What I'm not seeing is a range of articles from pundits saying how they got it all wrong, how Apple held to its position in spite of all of the external pressure to pivot, and delivered a great end result in spite of some unfortunate delays.
Of course, lets ignore all of the controversies and incredible cost of the AI development surrounding OpenAI, Microsoft, Google, and Meta. Let's not even talk about $1.1 trillion in unaccounted-for funds thrown away for the pursuit of the fantastical and made-up "superintelligence." No, Apple is failing, flailing, and can't keep its head above water.
Clearly. Apple is, and forever will be DOOMED™.
I didn't link all of it here, and perhaps I should do an omni post collecting it all, but if you didn't read my Apple AI reporting from 2023 to today, you should. Most of it was accurate, and what wasn't was due to poor reporting and assumptions made from external sources. For example, reports suggested multiple times that Apple would launch more extensions for Apple Intelligence like ChatGPT, and that led me to making a few assumptions that didn't come to pass (yet).
I will take a moment to pat myself on the back. This is my blog after all. Apple never swapped Siri's backend out for ChatGPT, Gemini, or any other system, which was always a ludicrous rumor I can't believe ever had any traction. I was beating that drum from the start even as, bafflingly, much of the Apple press went the opposite direction.
OpenAI's gambit
I shared all of this because it serves as the perfect background to OpenAI's supposed new product that was rumored this week. All while pundits continue to ask if Apple will ever succeed at AI, OpenAI is desperately searching for new revenue streams.
As every doom and gloom report arrived about Apple's supposed failures in AI, OpenAI dug itself deeper and deeper into a financial hole. Apple's lawsuit suggests it was OpenAI desperately seeking to learn some secret future plan as a key to extending its financial runway.
It seems a $300 to $400 portable donut-shaped object with moving parts, flashing lights, a microphone, and a speaker might be their best bet. That is, if Apple's lawsuit doesn't prevent it from launching.
Curious how much of OpenAI's product sounds similar to what has been rumored for Apple's AI hardware projects. Surely that's a coincidence, right?
I wrote previously about OpenAI swearing this wasn't a HomePod or like anything else on the market, then Mark Gurman did a lot of hair splitting in his story to continue that narrative. I'm not convinced, but John Gruber, linked above, did make a good point.
Perhaps calling this a smart speaker is wrong, I'll give it that. It sounds more like a toy or a pet, as Gruber said. I'm not so sure about a pet, unless you're thinking of Furby or that Talking Mario Flower.
However, it doesn't sound like it has a display, so it's voice first, and its primary way of sharing data back to the user will be speaking out loud. It may not be a literal HomePod or Echo Dot, but the interface sounds wildly similar.
Apple's Home Hub tablet device with a robotic articulating arm is closer to this rumored OpenAI product than a HomePod is, but they're in the same general vicinity. I'll give them some benefit of the doubt for now, but will remain skeptical of it being anything unique until I see it.
Besides, if this thing can't provide any functional value beyond what your smartphone can already do with ChatGPT installed, it will be dead on arrival.
It's a swing for the fences and reeks of desperation. A little robotic toy-like home product for over $300? I get it, that's what is ready to ship today, but if hardware is truly what will keep OpenAI from dissolution and future acquisition (My bet is on Microsoft), this ain't it.
I guess we'll just have to wait and see what Jony Ive's team has built when the smartphone-not-a-smartphone-wearable thing is revealed in the coming years.
But yes, Apple is most definitely the desperate one.
