Good Morning! A Miami developer made a $1B gain selling land to Ken Griffin for Carnegie Mellon's new Miami campus. Goldman's top 20 executives are set to unlock a special bonus worth $500M+ combined, with David Solomon taking home over $100M.

Starbucks is eyeing Chipotle in what could be the largest acquisition in restaurant history, and Vance says the US will suspend Microsoft from the green card program. Jeff Bezos says AI could enable three-day workweeks, and Tottenham's Xavi Simons enrolled at Harvard Business School while recovering from a torn ACL.

Plus: a Lincoln Park billionaire was accused of fueling a rat infestation in Chicago, and how to make friends in a new city.

Make that hefty NYC monthly rent start working for you. Earn points for it with the Bilt Card. Learn more about it here.

OpenAI has been planning one of the largest IPOs in history, which the company’s management recently pushed back to at least 2027 over AI safety concerns. But in addition to needing time to build some guardrails around new models, the company apparently might need time to find another $20 billion in revenue it had been mistakenly reporting.

Yesterday, new financial documents showed that OpenAI is approaching around $50 billion of annualized revenue through September, which sounds healthy but is about $20 billion below the $70 billion figure that had been widely reported just days earlier. 

And, even though OpenAI isn’t public, it’s still one of the most important companies in the world, and Wall Street noticed. AI-linked stocks sold off after the report, and Nvidia closed down 3%. Oracle closed down more than 5%. Some investors were already nervous about all the debt needed to finance the AI buildout, and finding out that OpenAI’s headline revenue run rate was materially lower gave them another reason to revisit the model.

The good news is that OpenAI didn’t exactly lose $20 billion. Part of the discrepancy comes down to how investors were trying to compare OpenAI with Anthropic. Anthropic includes revenue generated through cloud partners like AWS and Google Cloud in some of the run-rate figures investors use, and OpenAI doesn’t. So investors were effectively grossing up OpenAI’s number to make the two companies more comparable. But still, markets don’t love discovering that one of the most important numbers in AI needs a reconciliation.

And OpenAI is still losing money while committing enormous sums to chips, data centers, and computing capacity, and the entire AI trade increasingly rests on the assumption that revenue will grow quickly enough to justify hundreds of billions of dollars of infrastructure spending underneath it. So whether the industry’s flagship private company is running at $50 billion or $70 billion is not exactly an accounting detail.

Takeaway: $50 billion is still an absurd number. OpenAI generated a little over $20 billion of annualized revenue at the end of 2025, and the company is still growing at an absolutely insane pace. The problem is that when investors are discussing a $1.4 trillion valuation, “the business is growing insanely fast” and “the business is growing as fast as everyone thought” become two very different questions. 

Bilt is the rewards program for where you live that turns one of your biggest fixed costs - your rent or mortgage - into valuable points.

With each Bilt Card*, members can earn up to 1.25X points on housing payments wherever they live, with no added transaction fee. There is an option for every type of spender, ranging from a no annual fee card (see rates & fees) to a premium card offering 2X on everyday spending.

Bilt Points can be redeemed for travel with major partners (like United and Hyatt), used toward future housing payments, Lyft rides, Amazon.com purchases, and more.

Learn more about the cards to figure out which one is right for you.

Market Update

Economic Data

Fed Outlook

Movers & Shakers

Prediction Markets

Private Dealmaking

For more PE, VC & M&A deals, subscribe to our Buysiders newsletter.

Description:
Steven Pinker explores one of the most fascinating and underappreciated forces shaping human behavior: common knowledge, or the things we know that other people know, and that we know they know, and so on. Pinker shows how this seemingly abstract idea helps explain financial bubbles and crashes, political movements, social conventions, diplomacy, relationships, and even awkward everyday interactions. He combines psychology, economics, game theory, linguistics, and evolutionary science to explain why people sometimes openly acknowledge what everyone already knows, while at other times they go to extraordinary lengths to pretend that nobody knows. The result is a fascinating framework for understanding how shared awareness coordinates societies, creates collective action, and can suddenly cause entire groups of people to change their behavior.

Book Length: 364 pages
Release Date: September 23, 2025

Ideal For:
Investors, economists, entrepreneurs, psychologists, political thinkers, and anyone interested in behavioral finance, game theory, social dynamics, human psychology, and understanding why markets and societies sometimes move collectively.

“Knowing something is powerful. Knowing that everyone else knows it can change everything.”

Higher financing costs and tighter return thresholds have made entry price more consequential than ever. Buyers have less room for error, while bankers increasingly need to understand how far sponsors can stretch on valuation in a live process.

Mosaic’s new Reverse LBO helps answer that question quickly.

Enter a target IRR or MOIC, and Mosaic works backward to calculate the maximum purchase price, entry multiple, or bid premium that supports the required return. As growth, margins, leverage, financing costs, or other assumptions change, the answer updates automatically.

For bankers, that means a faster way to frame sponsor bid capacity, pressure-test valuation, and run scenarios during a process. For PE teams, it means quickly understanding how much they can pay while still preserving target returns.

See a Reverse LBO in action.

In 1973, UC Berkeley’s graduate programs admitted about 44 percent of male applicants and about 35 percent of female applicants. It looked like clear bias against women.

Then researchers split the data by department. In most departments, women were admitted at rates equal to or higher than men. Women had applied more often to the most competitive departments, where almost everyone gets rejected, and that dragged down their overall rate.

A trend can reverse once you split the data by a hidden variable. It happens in business constantly. A company’s blended margin can rise while every product’s margin falls, simply because the mix shifted toward the higher-margin product.

Before trusting any headline average, ask what is hiding inside it, and split the data by segment, cohort, or channel.

📣 Partner With Us: Get in front of an audience of over 1 million finance professionals, business leaders, and policy influencers. Submit a partnership inquiry.

📊 Wall Street Comp & Culture Data: Get the most detailed comp, carry, and culture insights across 1000+ Wall Street firms. Explore the data.

📈 Grow With Us: Work directly with the Overheard on Wall Street team to scale your finance brand. Schedule your free consult.

🧢 Wall Street Shop: Explore our collection of finance-themed apparel and merchandise. Visit the shop.

📬 Deals Newsletter – Buysiders: A curated roundup of major M&A, private equity, and VC activity. Plus access to private deal flow. Subscribe here.