Ben Bajarin

CEO & Principal Analyst @creativestrat. Full-stack tech analysis from silicon to markets. Advisor to tech leaders. Technologist at heart. Keeper of Bees.

[Ben Bajarin @BenBajarin]

Creative Strategies has been around for decades, but most of our work has lived behind client walls.

Diligence Stack changes that.

It brings our analyst work, industry access, market frameworks, and technology diligence into a public research product built for investors, while also serving the operators, strategists, and industry participants trying to understand where the next cycle of value creation is forming.

https://t.co/4dvQJtVaE8

[Ben Bajarin @BenBajarin]

@WilleeDubs @handleym99 @mweinbach At @mweinbach can explain for you.

[Ben Bajarin @BenBajarin]

Per Craig Federhigi on how much Google gemini stuff they use for Apple Intelligence:

"we don't have the Gemini app as our app. In fact, none of that client code is part of how we run an iOS for these models. We use none of the models that Google deploys to their customers, nor do we use the infrastructure and means by which they deploy models to their customers. And then, when it comes to the knowledge base, we, of course, don't use Google search or anything like that as the foundation of our system, so hope that's clear. This is the amount of the Google Assistant we use, which is none."

[Max Weinbach @mweinbach]

Using Apple Mail on macOS now

I just was able to reply to a phone OEM asking for my address by saying "reply to this email and send my address and number" in Spotlight and it did it in like 15 seconds.

Finally as fast as doing it myself

[Josh Elman @joshelman]

The technical details behind Apple Foundation Models are worth a read:

https://t.co/RI5qV9eKBU

[Ben Bajarin @BenBajarin]

@paurooteri Yeah all the others are theirs. They had multiple papers through the years explaining AFM. Cloud pro is Gemini base model.

[Ronald Mannak @ronaldmannak]

Really interesting. It’s an MoE model but instead of choosing experts per token, a small (and dense) model within the model selects which experts to use for the whole prompt. That way you can just load the experts in memory needed for the prompt.

[Max Weinbach @mweinbach]

The reason for Nvidia, Google, and Intel are the following:

- NVIDIA Confidential Computing with NVIDIA GPUs
- Intel CPUs with TDX
- Google's Titan chip

These are required to hit Apple's benchmark for AI inference off of Apple Silicon

[Ben Bajarin @BenBajarin]

@Teich50 @mweinbach Geopolitics where places don’t trust Google

[Ben Bajarin @BenBajarin]

And now Apple! Gotta monetize that infrastructure while poor internal teams still sit and wait for compute.

Need more compute!

[James Chanos @RealJimChanos]

So is $GOOG compute constrained as the deal w/SpaceX implied, but not so constrained that they can sell $35B of their chips to their biggest competitor (Anthropic)…?!

[Ben Bajarin @BenBajarin]

Can it share a number with 3 phones cause that’s what @mweinbach needs 😂

[MacRumors.com @MacRumors]

iOS 27 Tidbits: Share a Phone Number on Two iPhones, Independent Alarm Volume, Faster AirPlay and More https://t.co/ARDGshcK1u

[Marques Brownlee @MKBHD]

Tell me you have a foldable coming without telling me you have a foldable coming

[NVIDIA @nvidia]

Proud to collaborate on the next frontier of private AI. 🔒

Apple is expanding Private Cloud Compute to @GoogleCloud using NVIDIA GPUs.

https://t.co/6R6MP1SkVm

[Max Weinbach @mweinbach]

First impressions are new Siri is way more impressive than most think

[Ben Bajarin @BenBajarin]

@handleym99 Per @mweinbach it’s just the base model IP. Lots of additional training done by Apple to build it up.

[Ben Bajarin @BenBajarin]

Knock knock knock knock

Please let me in!

[Ben Bajarin @BenBajarin]

Podcast from WWDC in the books!

[Jukan @jukan05]

TSMC was too naive from the start.

Intel is emerging as a serious competitor. TSMC has been extremely strict about preventing Samsung-designed chips from being manufactured at TSMC, so why did it allow Intel to manufacture chips at TSMC?

At this point, TSMC obviously has no reason to allocate CPU capacity to Intel anymore. Instead, it will—and should—actively allocate wafers to Intel’s CPU competitors.

[signüll @signulll]

in today's keynote, apple produced this really interesting graphic that ironically outlines the core mechanics for a new type of operating system (for perhaps a new class of devices).

you can see how this moves the world from an app based ecosystem to an intent centric world.

i.e. you roughly do not need third party applications in this world at all esp when ai has the ability to construct & deconstruct interfaces / experiences on demand.

[Ben Bajarin @BenBajarin]

@minato71216 Apple trained the models using their data and only the cloud models using underlying base model of Gemini.

[Ben Bajarin @BenBajarin]

@azeem Downloading all betas to try so we will see!

[Ben Bajarin @BenBajarin]

Jensen talks a lot about useful AI. Hoping we see Apple demonstrate this for their ecosystem/platform/customers.

And really, I just want better incredibly capable search on mac and ios lol.

[Ben Bajarin @BenBajarin]

@tcrawford @Apple YOLO!

[Ben Bajarin @BenBajarin]

Who is with me and will put dev beta one on all your primary Apple devices tomorrow because YOLO?