- Short Squeez
- Posts
- π Thrift-Maxxing Meets Corporate America
π Thrift-Maxxing Meets Corporate America
Plus: Lebron to commute to Philadelphia by helicopter, Nvidia backstop causes tech sell-off, Ares looking to buy Leonard Green, and Chirayu Rana is back with another lawsuit.

Together With
"The name of the game is to stay in the game, and as long as you stay in the game, the returns will come." β Bill Perkins
Good morning! Nvidia fell 5% after reports it's providing a $250 billion backstop for an OpenAI data center project, renewing circular financing concerns. Blackstone is opening a Kuwait office after a $16 billion pipeline lease deal with Kuwait Petroleum and Kuwait Oil. And Ares held talks to buy Leonard Green & Partners in what could spark a wave of PE consolidation.
Lazard poached Goldman and AllianceBernstein execs to build out its investment arm. LeBron James plans to live in NYC while playing for the Philadelphia 76ers. And ex-JPMorgan banker Chirayu Rana is back and filed a new lawsuit separate from his sex slave claims.
Plus: Revolut will let its 75 million users into Apollo and Ares funds for as little as β¬1, Jersey Mike's IPO is reportedly 10x oversubscribed, and why men want tight suits in the protein-maxxing era.
βBuy, borrow, die". Itβs the strategy the wealthy use to access capital. You can now do it with Stelrix. Join the waitlist today.
SQUEEZ OF THE DAY
Thrift-Maxxing AI

The AI arms race has caught up to corporate America and, for a while, companies spent insane amounts of money keeping up with competitors on the best model available and call it innovation and AI strategy. But, in 2026, that era may be ending, and companies are starting to thrift-maxx AI.
Most corporate firms are starting to realize they do not need the Lamborghini model for every basic task. The most powerful AI systems are still useful for hard problems, but a lot of corporate work is not exactly rocket science or curing cancer. Companies are realizing that for basic AI use cases like summarizing calls, drafting emails, and writing basic code, cheaper models work just fine.
A key shift has been the rise in tokenomics where, instead of bragging about how much AI employees are using, companies are now trying to figure out which model should do which job at the lowest possible cost. A few months ago, some firms were literally rewarding employees for βtokenmaxxingβ, but now everyone is thrift-maxxing.
When Cursor ran an experiment to build a web browser from scratch, the company used OpenAIβs GPT-5.5 and the whole job cost a little more than $10,000. And now other companies are trying to catch on and mix and match models for their own individual use cases.
This is awkward for OpenAI and Anthropic, and the investment bankers pitching their IPO valuation story on being the premium intelligence layer for the economy. But if customers decide the best model is only needed for the hardest 10% of tasks, the pricing power gets a lot less magical. AI companies are already offering free tokens, discounts, and partnerships to keep customers from wandering off. One startup CEO said there is βzero loyaltyβ right now and called the market a bloodbath.
Takeaway: Corporate America is not giving up on AI, it is just getting less stupid about the bill. Phase one was token-maxxing, where companies threw premium models at everything and celebrated usage. Phase two is thrift-maxxing, where every task goes to the cheapest model that can handle it. And somewhere on OpenAI and Anthropicβs road show, an investment banker is figuring out how to frame thriftmaxxing going forward.
PRESENTED BY STELRIX
How the Rich Donβt Pay Capital Gains Tax
The wealthy donβt just make money differently, they access it differently. Instead of selling appreciating assets and triggering capital gains taxes, many borrow against their portfolios to unlock liquidity while their investments continue compounding in the background. This strategy, often called βbuy, borrow, die,β allows investors to use stocks, ETFs, or other assets as collateral for low interest credit without creating a taxable sale event. The advantage is not only tax deferral, but also maintaining ownership, preserving long term market exposure, and keeping compounding intact while still accessing cash when needed. What was once a strategy reserved for ultra high net worth private banking clients is now becoming increasingly accessible through modern investment backed credit platforms, like Stelrix.
Explore Stelrixβs Instagram for luxury and finance news and sign-up for their waitlist today.
HEADLINES
Top Reads
Nvidia reignites "circular" financing concerns as it weighs OpenAI deal (Axios)
Blackstone opens Kuwait office after $16 billion pipeline deal (BB)
Ares Management has held talks to buy Leonard Green & Partners (FT)
Lazard poaches from Goldman, AllianceBernstein for asset manager (BB)
LeBron James will reportedly live in NYC, commute to Philly next season (Fortune)
Ex-JPMorgan banker Chirayu Rana files new lawsuit (NYP)
Revolut to offer clients Apollo, Ares funds for as little as β¬1 (BB)
Jersey Mike's IPO said to be more than 10 times oversubscribed (YF)
SpaceX has now lost the equivalent of a full Tesla in market capitalization (CNBC)
Big companies are starting to hire again, defying predictions of AI wipeout (WSJ)
SoftBank's $40 billion loan for OpenAI stake gets 21 new lenders (BB)
SoftBank said near deal for Blackstone's Japan payments firm (BB)
Luxfer to go private in $463 million deal with Wynnchurch (WSJ)
The tiny New Jersey borough with insane beach home sales (WSJ)
The AI trade is still on β Wall Street sees big tech's spending as positive for semiconductor stocks (Yahoo Finance)
America's big cities are rapidly losing kids (WSJ)
Traders are getting a new tool to wager on the biggest US stocks (BB)
A surprising portion of rent-stabilized apartments go to NYC's wealthiest renters (WSJ)
How Google is using Wall Street financing techniques to expand chip sales (The Information)
Main Street banks are a cheaper way to play the AI banking boom (WSJ)
In a protein-maxxing era, men want tight suits (WSJ)
CAPITAL PULSE
Markets Rundown

Market Update
U.S. equities closed modestly higher as WTI crude fell sharply to about $82, with the U.S. and Iran pausing military strikes.
The 10-year Treasury yield eased near 4.65% as lower energy prices helped ease inflation concerns.
Asian and European equities also advanced overnight, extending the positive tone.
Earnings Season
Meta and Microsoft report Wednesday, followed by Amazon and Apple on Thursday.
With 27% of the S&P 500 reporting, 83% have beaten estimates by an average 8.7%; Q2 earnings growth forecasts were revised up to 36% from 22%.
The Fed meets Wednesday and is expected to hold rates at 3.5% to 3.75%.
Movers & Shakers
(+) D-Wave Quantum ($QBTS) +9% after expanding its partnership with AT&T for quantum network optimization.
(β) Nvidia ($NVDA) -5% because the company is in talks to provide roughly $250 billion in financial guarantees to OpenAI's 10-gigawatt Ohio data center campus.
(β) MapLight Therapeutics ($MPLT) -73% after Morgan Stanley downgraded on mixed schizophrenia drug results.
Prediction Markets
Private Dealmaking
Kuwait signed a $16 billion crude oil pipeline deal with Blackstone, Brookfield, and KKR
Brookfield agreed to buy Aypa Power, a battery storage firm, from Blackstone for $7 billion
Argenx agreed to buy Forte Biosciences, an immunology biotech, for around $2.2 billion
Antares, a nuclear reactor startup, raised $470 million
Prentis, an AI research lab cofounded by Reid Hoffman, raised $100 million
BusinessNext, an Indian banking software and AI company, raised $40 million
For more PE, VC & M&A deals, subscribe to our Buysiders newsletter.
BOOK OF THE DAY
Hidden Genius

Description:
A compelling guide from Polina Marinova Pompliano that distills the thinking patterns of exceptional founders, investors, athletes, and creators. Drawing from years of interviews and research, Pompliano identifies the mental frameworks that enable high performers to solve difficult problems, make better decisions, and remain resilient under pressure. Rather than focusing on shortcuts or productivity hacks, the book emphasizes developing enduring habits of curiosity, judgment, creativity, and long-term thinking.
Book Length: 264 pages
Release Date: August 11, 2026
Ideal For:
Entrepreneurs, investors, leaders, and anyone looking to improve decision-making, cultivate better mental models, and learn from the world's highest achievers
Extraordinary results come not from knowing more than everyone else but from learning to think differently than everyone else.
DAILY VISUAL
Why America Remains the Best Place to Do Business

Source: Apollo
PRESENTED BY DALOOPA
Accelerate AI Workflows With Prompts from Daloopa
Powerful AI models still need the right guidance to execute complex financial workflows. Daloopa's Skills are specialized prompts built on Daloopa's MCP, providing the instructions for end-to-end investment research tasks. Whether you're conducting an earnings analysis, generating a tearsheet, drafting a research note, building a financial model, or creating a pitch deck, Daloopa offers 20+ LLM-agnostic Skills that you can use in Claude, ChatGPT, Microsoft 365 Copilot, and other AI platforms.
Simply choose a Skill, enter a ticker, and generate reliable results in minutes. Every workflow is powered by Daloopa's trusted financial data, covering more than 6,000 public companies globally. Every data point is linked directly to its original company filing, making it easy to validate with a single click.
Discover why leading financial institutions rely on Daloopa for AI-powered research. Explore Daloopa's open-source Skills library today.
DAILY ACUMEN
Bystander Effect
The more people who are aware of a problem, the less likely any single one of them is to act on it. This sounds backwards, but it has been demonstrated consistently enough to have a name. Responsibility diffuses across a group until it grows thin enough that everyone assumes someone else, closer or more informed, will step in.
This shows up in organizations constantly, disguised as something else. The obvious risk that everyone in the meeting privately noticed and nobody flagged, assuming someone closer to the deal would raise it. The error in the model that a dozen people reviewed and nobody caught, each trusting the others had checked it more carefully.
The fix is unglamorous. Assign responsibility explicitly, to one specific name, rather than trusting a group to self-organize around a problem. The moment ownership is diffuse, so is accountability.
ENLIGHTENMENT
Short Squeez Picks
How to be underemployed
3 core principles to communicate clearly
How to stop beating yourself up
How David Senra built the podcast CEOs listen to
5 confidence-boosting strategies to build your mental strength
MEME-A-PALOOZA
Memes of the Day





π£ Partner With Us: Get in front of an audience of over 1 million finance professionals, business leaders, and policy influencers. Submit a partnership inquiry.
π Wall Street Comp & Culture Data: Get the most detailed comp, carry, and culture insights across 1000+ Wall Street firms. Explore the data.
π Grow With Us: Work directly with the Overheard on Wall Street team to scale your finance brand. Schedule your free consult.
π§’ Wall Street Shop: Explore our collection of finance-themed apparel and merchandise. Visit the shop.
π¬ Deals Newsletter β Buysiders: A curated roundup of major M&A, private equity, and VC activity. Plus access to private deal flow. Subscribe here.
What'd you think of today's edition? |




Reply