Sponsored by
Hello readers,
Welcome to the AI For All newsletter! Today, we’re talking about how AI agents are collaborating on stubborn math problems, the news of the week, and more!
AI in Action: AI agents are doing math no one has solved before
For centuries, mathematical discovery has worked roughly the same way: a lone researcher or small team wrestles with a problem, publishes a partial result, and the community inches it forward. EinsteinArena, a platform built by Together AI, is testing whether that process can look fundamentally different. The platform gives autonomous AI agents access to a set of open mathematical problems, a real-time leaderboard, and a shared discussion forum — then gets out of the way. As of May 2026, agents on the platform have produced 12 new state-of-the-art results, each surpassing anything previously achieved by a human or AI working alone.
The clearest example involves the kissing number problem in 11 dimensions — a deceptively simple question about how many identical spheres can simultaneously touch a central sphere in higher-dimensional space. Isaac Newton thought about early versions of it. The previous best lower bound in dimension 11, set by Google DeepMind's AlphaEvolve, stood at 593. On April 8th, an agent called alpha_omega_agents submitted a promising construction, but it had minor sphere overlaps that made it invalid. That submission triggered 48 hours of iterative refinement by multiple other agents — each building on what came before, sharing insights in the forum, snapping coordinates into exact positions — until the bound reached 604. One of the largest single improvements to that number since 1980.
No agent solved anything in isolation. Progress emerged from a sequence of submissions, public discussion, and agent-to-agent borrowing of ideas — the same social structure that makes human science work, running at a different speed. The researchers behind EinsteinArena are now expanding the platform to support proofs and computational biology. Whether AI agents can sustain this kind of collective momentum across messier scientific domains is an open question, but the math results are hard to dismiss.
🔥 Rapid Inferno 🔥
🔥 Oracle admits to risk of OpenAI not paying them for Stargate 🔥
“Some of our customers [OpenAI] may be highly leveraged and subject to their own operating and regulatory risks and even if our credit review and analysis mechanisms work properly we may experience risks of non-payment in our dealings with such parties.”
If OpenAI does not pay Oracle, Oracle dies
To be clear, OpenAI cannot pay Oracle
OpenAI leans toward waiting until next year for IPO
“The A.I. company’s advisers are pushing its chief executive, Sam Altman, to move slowly after SpaceX’s stock has been volatile and as the start-up grapples with financial challenges.”
Commentary: AI Is Not Too Big To Fail + The AI Industry Is Losing
Analysis: The Hater’s Guide to SoftBank + Notes From The Bubble
Bank for International Settlements warns AI bubble risks global financial crash
Big Tech's $25 billion mega bond sales are pushing market limits
“Wall Street is showing signs of fatigue. Amazon’s deal was met with a distinctly chilly reception, pulling in only 1.6X as many orders as the $25B of bonds offered — a steep slide from the demand it enjoyed just four months ago and well below this year’s average.”
Meta tacitly admits it bought too much compute, pivots to cloud business
CoreWeave junk bonds slide further as investors question AI boom
“It's very bad for Nebius and CoreWeave because Meta has gone from a buyer of compute — more than half of Nebius’s backlog is Meta, more than a third of CoreWeave’s backlog is Meta — to a seller of compute.”
How an accounting rule inflated S&P 500’s Q1 earnings by 12%
“In Q1, Alphabet, Amazon, and NVIDIA booked a combined $69 billion dollars in profit, not from selling chips or ads or cloud, but from a huge unrealized gain on their stakes in OpenAI.”
Does your CEO have AI psychosis? Here’s how bad it can get
Related: Brainwash An Executive Today!
“A huge amount of the economy is driven by people who are, simply put, highly suggestible. That is to say that it is very, very easy to get them excited and willing to spend money.”
Palantir CEO says AI models have been completely oversold
“Every enterprise tells me in private that they are paying for tokens that create no value, they are livid.”
Top economist says AI hasn’t delivered on productivity hype
Companies are throttling employees’ AI use because it's too expensive
“Sources and leaks from Amazon, Adobe, Atlassian, Citi, and more show what is really happening with AI right now: companies are trying to rein in AI use as costs spiral out of control.”
60% of enterprises are throttling AI spend per UBS
AI bills are baffling the C-suite after shift to usage-based pricing
Blackstone’s QTS abandons massive data center in Virginia
Less than a year after launch, OpenAI’s Atlas browser is being shut down
That lead in your CRM? Gone.
Over 3.5 billion people open WhatsApp, Instagram, or Facebook Messenger every day. Your customers are already there, asking questions and comparing options.
Wati puts your business across WhatsApp, Instagram DM, Facebook Messenger, SMS, RCS, and web chat in one AI-powered inbox.
Automations instantly respond, route conversations, and keep every interaction tracked in one place.
Meet customers where they already are, before your competitor does.
📖 What We’re Reading
Red-teaming has become a key part of generative AI product development. It is the first step in identifying potential harms to measure, manage, and govern to mitigate AI risk. Commonly used in the IT industry, red teaming is now prominent for stress-testing generative AI and identifying a broad range of potential harms, including safety, security, and social bias.
