Welcome, humans.
Well, it's official (sorta): OpenAI just "raised" $110B. We put "raised" in quotes because much of that headline number is letters of intent, not money in the bank. Here's how it actually breaks down:
Amazon: $50B total, but only $15B is deployed upfront. The remaining $35B is conditional on OpenAI hitting milestones (reportedly tied to its IPO or achieving AGI). Amazon also expanded its AWS deal by $100B over 8 years.
Nvidia: $30B, but much of this arrives as compute capacity commitments (3GW of inference, 2GW of training on Vera Rubin chips), not liquid cash.
SoftBank: $30B, described as a letter of intent that may or may not convert to actual capital by year's end.
So the real number? Probably closer to $15B in immediate deployable cash, with the rest gated behind milestones, hardware deliveries, and SoftBank CEO Masayoshi Son's mood. The round values OpenAI at $730B pre-money ($840B post), up from $500B in October. Supposedly it’s still open for more investors if ya wanna get in pre-IPO!
Now, one part of the above caught our attention: the AGI clause with Amazon. Last week, we heard a rumor that Microsoft and OpenAI's AGI deal might boil down to a surprisingly specific threshold: 50% of all productivity tasks that the current human workforce can perform, as judged by an independent panel.
Which makes you wonder: where is the benchmark that's supposed to track these "workforce tasks"? Personally, we thought that benchmark might be GDPVal, OpenAI's own eval that measures AI performance on real work across 44 occupations. Interestingly, GPT-5.2 has been sitting at 49.7% on the GDPVal leaderboard since last September.49.7%. The threshold is 50%. That's... close.
Could the upcoming GPT-5.4, reportedly set to release this week or next, be the first model to cross 50%+ on GDPVal? Is that why it’s been delayed the few times that it has already? And if it does come out this week, could it surpassing 50% on GDPVal qualify as "AGI" under Microsoft and OpenAI's deal? And IF it does... where's the independent committee that's supposed to make that call?
We think it's about time these companies come forward with the benchmarks and panels responsible for deciding all this. Otherwise, we're all left to speculate.
Here’s what happened in AI today:
The Pentagon blacklisted Anthropic, then OpenAI swooped in.
NVIDIA is building a dedicated inference chip using Groq's tech.
DeepSeek's new V4 model is set to drop on or before March 4.
Jeff Bezos's Project Prometheus AI lab wand to rollup firms disrupted by AI.
The Pentagon vs Anthropic vs …OpenAI??
Last Thursday, Anthropic drew a line: Claude cannot be used for mass surveillance of Americans, or for fully autonomous weapons. By Friday evening, President Trump ordered every federal agency to stop using the company's technology.
The Pentagon went further, designating Anthropic a "supply-chain risk to national security"... a label normally reserved for companies deemed foreign adversaries to the US, like Huawei. Anthropic has vowed to challenge the designation in court, calling it "retaliatory and unprecedented."
The backstory: Anthropic had a $200M Pentagon contract, and Claude was the only AI model on the military's classified networks. The Department of War wanted "all lawful use" access. Anthropic said fine, except for two things:
No mass surveillance: Don't use Claude to collect Americans' geolocation, browsing, and financial data from data brokers.
No autonomous weapons: Keep a human in the loop before anything fires.
The DoW's "compromise"? According to Axios, the contract language they sent overnight included loopholes that would let those safeguards be overridden at will. Anthropic CEO Dario Amodei responded: "We cannot in good conscience accede to their request."
Needless to say, the White House did NOT like this…
Now here’s the ultimate twist:
Hours after the government moved against Claude…
And after 300+ Google employees and 60+ OpenAI staffers signed an open letter urging their companies to hold the same line…
And after Sam Altman went on TV and said it has the same redlines as Anthropic…
…OpenAI announced its own Pentagon deal. Now, Sam Altman claims this deal includes the same two red lines plus a third (no AI-powered social credit systems).
But Altman's announcement on X got community-noted almost immediately; first of all, government officials contradicted his framing, saying OpenAI will actually allow "all lawful purposes." Altman held an AMA the next day; if you want to read his answers to the specific concerns that got voiced, go check it out!
Why this matters: hours after the ban, the US military proceeded to use Claude during air strikes on Iran. CENTCOM had it running intelligence assessments, target identification, and combat simulations. So the gov using Claude in real combat scenarios is not a hypothetical scenario; it’s actively happening, right now.
As MIT physicist Max Tegmark put it: the AI industry spent years lobbying against regulation, promising to govern itself. Now “we have less regulation on AI systems in America than on sandwiches.” This was by design, as the AI industry has done the opposite of regulatory capture (where the winners in an industry lobby the government to regulate downwards so competitor upstarts can’t comply): they captured the government to avoid regulation entirely.
Every AI company will eventually face this question: when the government says "give us everything," do you comply or push back? We posit an interesting thought experiment: If a person can declare themselves a conscientious objector to being used by their country in war…. could a company not do the same?
FROM OUR PARTNERS
The $4 Billion Problem Hiding in Every Fast-Food Location
You want a quick bite, but the line’s too long. So you bail. The top frustration for 93% of monthly fast-food visitors, long lines equal missed meals and missed revenue. So brands like White Castle use Miso Robotics. Miso’s AI-powered restaurant kitchen robot works 2X faster than average fry cooks. That means shorter lines and up to 3X more profit per location.
This is a paid advertisement for Miso Robotics’ Regulation A offering. Please read the offering circular at invest.misorobotics.com.
Prompt Tip of the Day
Recently, somebody asked why we’re so focused on Anthropic/Claude in the Prompt Tip of the Day lately. It’s a good question!
Frankly, it's because no one else is innovating as much as Anthropic for general AI use-cases right now, and you’re probably behind if you're not learning how to use Anthropic’s tools (maybe Perplexity Cloud changes that; still TBD).
I know that sounds dramatic, but neither Google or OpenAI are making as easy to use, practically useful, truly agentic tools for normies (like us). Where is OpenAI’s answer to Cowork, their supposed Google Workspace competitor? We heard a rumor it could be coming in the next few weeks… but they’re sleeping on it.
It feels increasingly like the trick to working with AI is reducing the total number of prompts you need to make in the first place, through actual automation.
Things like plugins, and skills, and now scheduled automations build those automations into the system for you. Worry less about prompting, and more about building systems.
So yeah, today’s prompt tip is this: time to learn how to effectively use skills!
Two resources will help you do this:
Anthropic’s newly released Skills explainer video (see below).
This 32 page guide, which we wrote a TL;DR on here (we recommend all of the above!)
P.S: in that same spirit, should we retire the “Prompt Tip of the Day” branding and rebrand this section something like “AI Skill of the Day” so it’s more broadly applicable to what you ACTUALLY need to learn in 2026? Let us know!
Rebrand PTotD? Yay or nay?Choose your favorite rebrand option below (if the majority selects "Keep Prompt Tip of the DAY", then obviously we won't rebrand) |
Treats to Try
*Asterisk = from our partners (only the first one!). Advertise to 650K readers here!
*Stop typing long prompts. Dictate full context and Wispr Flow turns it into a clean, structured prompt you can paste into your AI assistant in seconds.
Auto-memory lets Claude Code remember your build commands, debugging patterns, code style preferences, and project architecture across sessions so you don't have to repeat yourself.
Claude Import Memory lets you switch to Claude from ChatGPT or other assistants without starting over; it imports your conversation history and preferences.
Now I Get It! helps you upload any scientific PDF and get an interactive web page explaining it in plain language.
Google AI Edge Gallery runs AI models offline on your iPhone for chatting, image queries, transcription, and mini-games.
Pixel auto-builds creatives, audiences, and campaigns across LinkedIn, Meta, Google, and X for you based on a simple description or webpage link.
Notra turns your daily work (meetings, Slack, docs) into publish-ready content like LinkedIn posts and blog drafts.
Orca Engine lets you play, mod, and host Minecraft servers in your browser by chatting with AI to install mods, generate datapacks, and troubleshoot.
Around the Horn
NVIDIA is building a dedicated inference chip using Groq's technology, debuting at GTC next month. OpenAI committed to 3GW of capacity as the chip's biggest customer.
OpenAI was sued by Kate Fox after her husband's ChatGPT dependency led to psychosis and suicide.
A consequence of the Pentagon vs Claude vs OpenAI saga: Claude beat ChatGPT in US App Store downloads after the Pentagon blacklisted Anthropic. Katy Perry posted a heart over Claude's Pro plan. Secretary Hegseth: Chief of Claude Marketing??
Goldman Sachs flagged specific AI-resistant “Halo” stocks driving UK / EU markets to record highs.
Howard Marks published a memo arguing AI has hit "Level 3" autonomous agents replacing knowledge work.
DeepSeek’s new V4 model is set to be released on or before March 4, timed to correspond with China’s “Two Sessions” (the country’s biggest annual political event).
China shipped 36x more humanoid robots than US competitors, explaining why it has won the early market.
NFL staffers voiced fears that AI could eliminate scouting and quality control jobs.
Jeff Bezos's "Project Prometheus" AI lab reportedly seeked tens of billions to acquire companies being disrupted by AI.
FROM OUR PARTNERS
Coresignal's Data Search enables non-technical professionals to create high-quality B2B lists and datasets in a few simple steps:
Type your search in plain English
Get a preview of up to 100 matched records
Enrich selected or all records instantly
Generate a dataset of up to 10,000 records in JSON
The result is a clean, structured company or employee list that is ready for CRM import, outreach, analytics, AI platform development, or internal tools.
 The identities of these men have been wiped for their own protection (lol jk) | That’s all for now.
|
