Bill Donahue on Needham: Why the AI buildout hasn't paid off yet - 'I think that’s probably what’s actually been going on, David, as opposed to Laura Martin’s explanation. The testing began, and most CEOs are discovering that AI investment is currently not valuable, and future or on-going significant AI investment is on the chopping block, as opposed to employees.'
on Google's breakthrough is Apple's gain - 'Apple’s Mixture of Experts allows one to effectively use models that are 1/10th the size of the entire LLM. If Goggle’s can independently reduce the size of models in memory by 6-fold, then we could enter a period when a working, functional LLM could be resident on a device, just as Apple has been working towards. Of course, the speed of this is dependent on the silicon on device and only Apple has the chips to do this.'
on Google's breakthrough is Apple's gain - 'Hmm, Perhaps Apple, working with Google on AI, saw this coming. That may have factored into negotiations with memory suppliers, who don’t want to be caught with massive CAPEX to build manufacturing capacity for demand that may slacken. That may explain Neo’s aggressive pricing.'
on Needham: Why the AI buildout hasn't paid off yet - 'The other side of the coin for a CEO is the risk equation. Employees are a known risk relative to their expense. AI might be cheaper, but what if an AI gets it badly wrong? What’t the likelihood of that? Is there a way to evaluate that risk prior to firing employees and ‘hiring’ AI?'
on Google's breakthrough is Apple's gain - 'It’s early in the game and things are changing fast and furious.'
on Needham: Why the AI buildout hasn't paid off yet - 'I believe that part of the friction is indeed a human element, but not necessarily an emotional connection. Some human jobs, or jobs done by humans require some sort of human element, and not just processing data and spitting out an answer or applying requests to algorithms. For example, if you have ever called a medical office with a complicated situation, and went round and round with a bot while thinking if only I could talk to a human, this would be so much easier. I believe the above graph is consistent with my assumption regarding the need, sometimes, for the human element.'
on Needham: Why the AI buildout hasn't paid off yet - 'Theoretical AI coverage = venture capital ROI enrichment x infinity. No introspection required.'
on Is this why Apple jumped Thursday morning? - 'A nice “little” bump so far. But similar bumps have been falling back. Still in that under 260 trading range. So I don’t give it much credit until it breaks through.'
on Premarket: Apple is red - 'Apple is off $0.68 or 0.27% at $251.94 pre-market while the major market indexes are also in the red before the bell. I’m expecting the S&P 500 above 7,600 by year-end (it closed yesterday at 6,591.90). I see the current market tumult as a long-term buy opportunity. This buy opportunity includes Apple in its current trading range.'
