Most companies fail because they make bad calls. Google is failing because it made the logical ones. That’s the twist the creator of this video breaks down, and it flipped how I think about Google’s whole situation.
Here’s the setup. A year and a half ago Google was on top, shipping the most powerful model around. Now Jeff Dean is leaving to start his own company, Demis Hassabis just stepped down as DeepMind CEO, and people are openly asking what Google is even doing. The video creator connects the dots, and the story underneath is wilder than the headlines.
🧭 The old way vs the new way
The author points to something Google fans forget. Google wrote “Attention Is All You Need” back in 2017, the paper every modern AI system is built on. OpenAI, Anthropic, ChatGPT, all of it traces back to that research. Google published it openly and then sat on it.
Even crazier: a former DeepMind engineer (now a Codex lead at OpenAI) says they had a ChatGPT-style product a full year before ChatGPT shipped. It was ready. So why did nothing happen?
The old-way thinking: protect the cash cow. Search plus ads is maybe the best business ever built, so anything that might disrupt it gets blocked. The video creator quotes the claim directly, that DeepMind was blocked from shipping products that could threaten Search. That’s the trap.
🪤 The innovator’s dilemma
This is the core idea the author leans on. Well-run companies that listen to customers and invest carefully can still lose, because the smart short-term move becomes the long-term mistake. Customers wanted more search results, so Google gave them more search results. Meanwhile large language models grew up on the outside and started eating the whole model of ten blue links.
I found this part genuinely clarifying. It’s not that Google’s people were dumb. They were optimizing for the wrong horizon.
The post’s author reads the departures through that same lens:
- Hassabis wants to think long term (curing disease with Isomorphic), not babysit a business chained to quarterly earnings.
- Jeff Dean reportedly had unlimited budget and compute inside Google and still felt he couldn’t build his vision there. He left to start Discovery Loop.
- Middle management that likes things exactly as they are gets blamed for part of the culture drag (the creator flags this as his own speculation).
💡 Why the author is still bullish
Here’s the pattern break. Despite all the bad news, this industry pro argues Google is far from finished, and lists why:
- Data. Mountains of proprietary training data nobody else has.
- Chips. Their own TPUs, now eighth generation, so they’re not just fighting for Nvidia GPUs.
- Hardware reach. Android powers much of the world’s phones.
- Cash. A profit machine that funds huge bets.
His boldest take: stop pretending to fight at the absolute frontier and go all-in on open source instead. Build the best open model on the planet, get everyone building on it, then sell the TPUs that run those models most efficiently. As the creator puts it, if open source proliferates and Google serves the hardware behind it, they win either way. They already have the Gemma family to build from.
What stuck with me is how calm the conclusion is. Losing your best people can make a team stronger, and Google still holds a stacked deck.
Want the full breakdown, including the quotes and the pitch-deck detail that made me laugh? Watch the whole video, it’s worth your time.