Google is desperately trying to find their place in this brave new world, where the only thing they make well is hosting other people's VMs and LLMs. It is so sad. It was their labs that invented the transformer, yet they failed to capitalize on it. Sure, many people are in a position to have to use Gemini via APIs, myself including, and not happy about it... because it is dirt cheap compared to most all, but still this is not what I would call AI-era domination or even stable presence.

Meanwhile Microsoft and Amazon is eating their cake with Azure and AWS. A whole new generation of smart kids is now starting their day with ChatGPT (not me, I do Claude, but same point), instead of Google... so for many people it is not Facebook nor Google Search where "the internet starts from". This is massive loss. Broadcom is probably going to eat at their push for in-house TPUs. And surely Apple already ate everyone's cake on the affordable laptop with Neo, which is incredible for Apple to do, as they've always been roughly 1.5 the price of windows competitors. And Apple did it years after Google forayed into Chromebook, which this Gemini Laptop basically is version 2.0 of. The moment ads start showing intertwined with GPT output, Google is roast, most their revenue is ad-based, no matter the very strong (and perhaps top) cloud tier and Android licenses. Regarding this - Fuchsia is still not used in any frontier product, and poor Android is so much Java-tied, that it basically lives in the 90s from in a certain sense.

We can only speculate what the reason for this all is, but I'll put a very unpopular bet on Google shrinking massively the next few years into something aggressively robotized which looks more an utility company, than a SOTA research camp. For what is worth IBM was clairvoyant enough to do this shrinking years ago, and now can brave for thousands of COBOL lines being rewritten, and then some layoffs. But IBM has evolved an ecosystem that keeps companies locked in many dimensions, similar to Oracle perhaps, while Google does not.

I wonder what's with the moonshots side of the business of Google and could they scale that.

They are still very strong in versatile r&d. And AI would open many new opportunities.