Why Google May Win the Long Game in AI Dominance

if you can't read the medium article
https://archive.ph/tSupQ

I was just about to ask if someone could download the PDF!

thank you

Awesome article - thanks for sharing. I'm much more bullish on Google over GPT after reading that which I didn't think would be the case.

I don't love how much they poo-poo'd on Microsoft. Not sure what direction they will go (continue to rely on OpenAI or try and build their own frontier model) and I get why they did given they don't actually have a product here like Google. That said,
• They already have all the enterprise-relationships and arguably the greatest enterprise sales team on the planet.
• They have a massive research team that I'm *confident is doing a lot of work in secrecy to try and enter this space. *I haven't worked there in ~5 years, and that's not based on any insider information or concrete evidence
• They can get rapid enterprise adoption even on inferior products through clever accounting & leveraging executive relationships. They just roll their next big product in existing enterprise contracts while discounting their core products to drive adoption and reliance and win market share.
• I'm a big believer that we're very close to diminishing returns in terms of intellegence/performance. Not like "we hit a ceiling" but more so models can keep getting smarter but most users won't notice much of a difference...
• We've already seen it's been relatively easy for competitors to catch up to the frontier models with so much research out there. Hard to discover new and innovative solutions to building LLMs but easy to replicate.

Not too late to buy more GOOGL :joy:

A few things to watch out for:
1. OpenAI's chip design with Broadcom (they need to drop COGS, and yeah Google's got TPU's)
2. Sora 2 (Sorry, Google never cracked social)
3. Even more "partnerships" OpenAI has with other companies (the too big to fail argument is worth monitoring)

I think the article undersold Google's advertising capabilities. The second they want to turn on ads in all the modes, they have the marketplace and distribution. OpenAI has to build that from scratch

I am more convinced everyday that there will be no winner takes all scenario. And I think that is the best possible outcome not just for a healthy market but given the nature of AI, for humanity as well.

Agreed, though I wouldn't be surprised if Google ends up > 70%. They have the full stack from data centers to chip design to more personal data than anyone, plus the best means in the world to monetize it. OpenAI and Anthropic have to pay margins to other companies at each of those tiers

OpenAI has the same pain as Facebook. They don’t own the device.

It would be hilarious if FB buys OpenAI

(Somehow)

> Sora 2 (Sorry, Google never cracked social)
For sure, Sora was part of the main storyline in the latest South Park. I've seen nano banana get some minimal play recently though. The real question is if/when they bring YouTube into the mix, which could amp up the usage

Yep Google’s vertical integration is a huge advantage

But ChatGPT is so popular in the mainstream it could end up owning the name of a LLMs like Kleenex/tissues

Exactly, How is OpenAI going to leap past 20 years of Google building the knowledge graph.. It can’t.

OpenAI is already streaming terrible ads geared toward college students, but it still feels like an overvalued B2B saas company to me. I do not think there is a mainstream use for it without integration. We already have google assistant and alexa so i dunno what the play is for them. We hear the kids mostly like ai to cheat, not to be creative so here we go :popcorn:

I'm teaching my kids how to use AI to generate/create. They're mainly doing world-building exercises.

So whereas where I might've been a forum moderator for an MUD based on Legend of Zelda - I have one kid that strait up spins up fan-fiction of a popular pre-teen book series, introduces her own characters, and explores what happens next in the world she's building/immersed inside of ChatGPT.

So instead of playing "make pretend/stop-motion videos" in their own imagination, they're thinking about making that happen with more precise words and generating that in Veo 3 or Sora 2.

Occasionally there's a filter inside Google Learn About, but kids being kids, they'll always find off-label uses. For example, instead of going to dumb browser-based games (e.g. Newgrounds or whatever) they can just prompt a game into existence.

It's weird out there and not all use cases were inspired by dad.

I don't under estimate it in the long term, but disappointed with Gemini so far.

Any AI where I have to copy and paste is behind the curve. Being able to use mcp in ChatGPT and Claude make them my daily tools.

It would have been nice to see them add access to mcp and/or other Google products.