Figure went from a $2.6 billion valuation in early 2024 to $39 billion in 2025. This is what the trial and error behind that jump looks like. Nothing about it happened overnight.
Faraday Future demoed its $89K humanoid in Brooklyn Bridge Park on Sunday. It wasnât autonomous: CTO Hong Rao teleoperated it by remote while it walked, danced and sang to visitors, and the team collected real-world interaction data. The long-term goal is roles like babysitting, teaching and companionship.
Aigen says more than 100 of its solar-powered AI farm robots have logged over 15,000 hours of autonomous fieldwork, pulling weeds without herbicides. Thatâs 15,000 hours with zero âsorry, you cut outâ moments, which is more than my last Zoom call managed in 15 minutes.
Eka Robotics raised $150M at a $1.45B valuation. It says its gripper picks about 1,500 objects an hour vs roughly 600 for a person, and it can lift a raspberry without crushing it. Gentler than the self-checkout yelling âunexpected item in the bagging areaâ at your raspberries.
In September, Googleâs fix was simple: move the traffic onto paths that still worked. Thatâs the lesson from all three stories. You need a second path that doesnât depend on the first, before you need it. Most homes and offices have exactly one internet connection. If Googleâs backups can fail, a setup with no backup is just waiting its turn. 8/8
The plan was to announce the bugs and the fix together on Thursday. Then the details leaked online. PHP rushed the patch out a day early, and within about half an hour, attackers were already trying the bugs on real servers. Web admins spent the day updating and restarting every server they had, racing people who already knew where to look. 5/8
This wasnât Googleâs first bad day this fall. On September 1, during routine maintenance in Iowa, a technician unplugged every fiber cable on a set of routers, one after another, in 13 minutes. Google builds those connections with multiple paths and separate power so no single failure can take them down. One person with a cable list did it anyway, and part of Google Cloud was offline for four hours. 6/8
Cloud regions are split into zones so that when one fails, the others keep things running. But redundancy only helps if you actually use it. Any service that lived entirely in that one London zone went down with it. As one expert put it afterward, a backup is only useful if it doesnât rely on the thing that just broke. 3/8
Then on Wednesday, PHP had its own emergency. PHP runs a huge share of the websites you use every day, and researchers using AI tools found two bugs that let attackers run their own code on unpatched servers without logging in. One security outlet called them among the most severe in software history. Theyâd been sitting in PHP for more than a decade. 4/8
In one week, a power failure in a single Google data center knocked UK banks offline, and two critical PHP bugs leaked before the fix was ready, putting a huge share of the web in a race against hackers. Both stories come down to the same problem, and Google already had a warning about it in September. đ§” 1/8
On Tuesday morning, a âpower eventâ hit one zone of Google Cloudâs London region. Revolut customers couldnât pay by card, pull cash from ATMs, or even log into the app. Lloyds, HSBC, Royal Mail, Rightmove and parts of the NHS had problems too. It took until midday to fully recover. One building lost power and a whole country found out what its bank runs on. 2/8

