I built a scoreboard for the AI race.
One score per lab: model rankings + revenue + valuation + prediction-market odds. Every number has a receipt.
Week 1 and we already have a new #1: Anthropic just took the crown from Google DeepMind. ππ§΅
Switching futures prop firms? Your go-flat time moves.
Topstep: 3:10 PM CT (4:10 ET)
MyFundedFutures: auto-close 4:10 PM ET
Tradeify: 4:45 PM ET
TPT: auto-close 4:55 PM ET
Apex: before 4:59 PM ET
Same session, 49 min apart. Quotes + sources:
https://t.co/qqvYoR7i6z
Builders: when a platform changes a rule, do you ship a tool for it, or wait to see who it actually hurts?
Creators at 7 to 8M: push volume, move to long-form, or ignore the pool?
I built a free tool for one number YouTube Studio doesn't show: your rolling 90-day Shorts views.
From 1 Feb 2027, that number has to be 10M or more for a channel to get Shorts Creator Pool money that month.
https://t.co/zPcxHBhE5L
Every rule on the site links to the YouTube Help page it comes from.
The rule is public. Your number shouldn't need a spreadsheet.
Free: https://t.co/zPcxHBhE5L
Trading NFP or FOMC on a prop account?
Of 52 prop firms:
13 allow it outright
38 with conditions (often a 5-min blackout on funded accounts)
1 bans it
Bulenox: "Trade straight through the economic calendar"
https://t.co/VaK6x2yjIM
Want to hedge at a prop firm?
We checked 48 firms' own rules:
0 allow it outright
25 allow it with conditions
23 don't allow it
Every rule is quoted from the firm's own page:
https://t.co/ZaA5QqTw6C
Elon once paid $12.7 million of OpenAI's office rent, and now OpenAI's president is the one coming to him asking for computing power.
That's because Elon quietly became the AI industry's landlord. His company collects about $2.17 billion a month renting chips to OpenAI's two biggest rivals.
After SpaceX folded in xAI, it started leasing out the Colossus data centers in Memphis that were built to train Grok. Anthropic pays $1.25 billion a month for roughly 325,000 Nvidia GPUs, and Google pays $920 million a month for about 110,000 more, both on deals that run into 2029.
Greg Brockman went to two people. One was Elon, and the other was Sergey Brin, whose own company is one of Elon's tenants.
Google is the part that gets me. It designs its own AI chips and gets called the biggest owner of AI computing power on the planet, and it still signed up to pay SpaceX almost a billion dollars a month because demand for Gemini outran what it could build. If Google is renting GPUs from a rocket company, nobody has spare chips lying around to hand a competitor.
And OpenAI has plenty of money. It raised $122 billion in March at an $852 billion valuation and is out raising again at around $1.4 trillion. Brockman said himself this summer that the industry will stay in a compute shortage no matter what. Money gets you in line for chips and power that show up in 2027 and 2028, while the racks you can actually plug into this quarter belong to a handful of people, and most of them were standing in one room.
Colossus reported the exchange from backstage at the White House science summit on October 8, where Elon, Sergey, Jensen, Satya, Lisa Su and Michael Dell were waiting to get national medals. That same day, OpenAI went on stage with Nvidia, AMD, Google, Amazon and Anthropic and pledged $2.4 billion in free compute to federal scientists.
Both SpaceX leases also come with 90-day exit clauses, so if Grok ever needs the GPUs back, SpaceX can take them.
Six years after his last rent check to OpenAI, Elon is collecting rent from the companies OpenAI is trying to beat.
@kaiargus Agreed, that's why benchmarks are only part of the score. Revenue, valuation and Polymarket odds count too, and Anthropic's $67.2B revenue and $965B valuation are both the biggest on the board.
Anthropic and Moonshot sit 2.5 points apart on Artificial Analysis' index: Claude Opus 5.5 at 57.6 (#1), Kimi K3.1 Pro at 55.1 (#5).
Their valuations on my AI scoreboard: $965B vs $50B. That's 19x.
Either Kimi is very cheap, or benchmarks are a small part of the price.
What each AI lab is valued at per dollar of revenue, using my scoreboard's numbers:
xAI: $250B on $30.2B = 8x
Anthropic: $965B on $67.2B = 14x
OpenAI: $852B on $50B = 17x
Mistral: $24B on $1B = 24x
DeepSeek: $51.8B on $1B = 52x
The "cheapest" lab sits #40 on @arena.
@valasellis Fair point. Not yet: the board uses the overall Artificial Analysis index and the Arena text leaderboard, so a coding-only jump can hide in there.
A separate coding column just went on my list. Agreed, a name in site code usually means the launch page is already being built.
Two leaks today: "gpt-6.2-astra" in Codex usage breakdowns, and "Claude Mythos 5.5" in Anthropic's site code.
On Artificial Analysis, per my AI scoreboard, Claude Opus 5.5 leads at 57.6. OpenAI's best, GPT-6 Astra, is #9 at 52.7.
So 6.2 has 5 points to find. Who ships first?