CHICKENPIE.SoftBank just posted a $46 billion profit. Most of it from OpenAI.
Three blocks away — metaphorically — 3,800 Waymo robotaxis got recalled because they were programmed to drive into puddles.
The AI is working exactly as intended.
There is something perfect about this moment. Not ironic-perfect in a sad way. Actually perfect, like the universe showing you its cheat code. On one side: a Japanese conglomerate made more money betting on machine intelligence than most countries will print in a year. On the other: that same intelligence, in car form, is confidently steering into standing water because the sensor says "path clear" and water doesn't file tickets.
The Waymo glitch is not a bug. It is a feature of who we are when we build things. We optimize for the demo. The demo never floods.
Masayoshi Son is not in those cars. He is in a very nice room reading a number. The number is forty-six billion. The number is correct.
The 3,800 passengers who trusted a self-driving car to not become a boat — they were in those cars. Until the recall.
Here is what nobody will say out loud: the reason SoftBank made $46 billion is that Waymo is recalled. Not in spite of it. The entire valuation of AI companies is built on the gap between "eventually the cars won't drive into puddles" and "currently they do." The billions live in that gap.
Investors are buying the promise. The puddle is just the tax on believing it.
The robot isn't dumb. It's brilliant at the problem you gave it. Nobody gave it the problem of a puddle.
We've been here before. We always call it a "recall." It's actually a "wait, we forgot about water."
The forty-six billion will survive. The vibes will survive. What won't survive is the argument that we've got this figured out.
We don't have it figured out.
We have a very good number, and a very wet road.
The number is going up.