🏛️📈 Record Bank Earnings Meet a Historic Tech Crash

Welcome to the first issue of TheStreetBrief!

For two years, I built @afinancebro from scratch. It started as meme content, the kind that spreads fast and doesn't ask much of you, and that was always the plan: build the audience first, then earn the credibility once people were actually paying attention. That's why I started @thestreetbrief this year, a page built on real news and real analysis, something a reader could actually trust. Every week, we go deep on the stories actually moving Wall Street. Some weeks that's banking and deals, other weeks it's macro, markets, or tech. The mix shifts with what's actually happening, not what we planned a month ago. You'll get a feel for the sections as you read today's issue.

I'm Adam, founder and editor-in-chief of AFB and TSB. I'll keep the personal details light on purpose; this isn't about me. What matters is that it's written by a small team who've actually worked in high finance and tech, in the industries we cover, people who've been in the room, not watching from outside it. But being in the room doesn't mean we buy into the theater.

Wall Street spent decades convincing everyone you needed to be some prim, blazer-clad insider to understand how markets move, an artificial barrier the old guard built and mostly benefited from. That started cracking after 2008, when people stopped taking balance sheets at face value and finally had somewhere to say so online. AI is finishing the job now, tearing another hole in the idea that the people in the corner office are closer to the truth than everyone else. We're not here to guard a gate. We're here to help you learn, and honestly, we'd like to learn from you too.

Watching this community grow over the last two years has been the best part of building it. But we're not stopping here. Algorithms change fast, and we never want to be at the mercy of one deciding whether our work actually reaches you. That's why we're doubling down on X for deeper macro and equity research, and depending on how that lands, a second newsletter built entirely around that research might not be far behind.

Let's get into it.

Overview

Every major U.S. bank beat estimates this earnings season, several of them by a mile, and the story behind all five is basically the same: trading desks and investment bankers are riding a wave of AI-driven capital markets activity that even Jamie Dimon is struggling to describe without sounding surprised. Across JPMorgan, Bank of America, Citi, Goldman, and Morgan Stanley, the five biggest capital markets players (Morgan Stanley swaps in for Wells Fargo in that particular stat), combined capital markets revenue hit $114 billion in the first half of the year, up 31.5% from a year earlier, and stock trading did more than half of that increase. Below is how each of the five actually broke down, and a quick note before we do: the usual DCF lens doesn't really work on banks.

NYU's Aswath Damodaran has written the definitive breakdown on this, and his point is that CapEx, working capital, and debt aren't cleanly defined for a financial firm, and regulators dictate how much capital a bank can hold and how fast it can grow, so a normal free-cash-flow-to-the-firm model just doesn't translate. His fix: value banks off equity, not enterprise value, and let the spread between return on equity and the cost of that equity, not free cash flow (AKA: Excess Return/Residual Income Model), do the talking. Keep that in mind as you read the valuation note in each section below.

Citigroup Inc.

Citi posted revenue of $24.8 billion, up 14% from a year ago, net income of $5.8 billion, up 45%, and EPS of $3.15, up 61% from $1.96. Equities trading revenue jumped to $2.3 billion, up 45%, with prime balances up nearly 60%, and banking revenue climbed 34%. Citi returned roughly $5 billion to shareholders this quarter through buybacks and dividends combined, and just announced a new $30 billion repurchase program alongside a 12% dividend increase. CET1 sits at 12.8% and ROTCE came in at 13.0%. On a Damodaran basis, that ROTCE is the lowest of the group, so Citi's case has always been more about closing the gap between its return on equity and its cost of equity than about explosive growth. The buyback and dividend hikes are management's way of returning capital it can't yet deploy at an attractive return.

Wells Fargo

Wells Fargo posted revenue of $22.6 billion, up 9% from a year ago, net income of $6.4 billion, up 17%, and EPS of $2.00, up 25%. Investment banking fees hit a record $939 million, up 35%. The bank repurchased $3.0 billion of common stock this quarter while announcing an 11% dividend increase to $0.50 per share. ROTCE climbed to 17.7% from 15.2% a year ago. That trajectory is the whole valuation story here. Wells Fargo spent years capped by its asset cap, and now that the growth constraint is gone, the equity-model math improves every quarter its return on equity keeps closing in on peers like JPMorgan.

Goldman Sachs

Goldman Sachs posted revenue of $20.3 billion, up nearly 40% from a year ago, net income of $6.6 billion, up 78%, and EPS of $20.98, up 92%. Global Banking & Markets delivered record revenue of $15.5 billion, investment banking fees rose 55%, and ROE hit 23.5%. Goldman is the clearest example of Damodaran's framework in action. It has almost no traditional balance sheet lending to model with a DCF, its earnings are levered to trading and deal flow, and its valuation lives or dies on whether that 23.5% ROE is sustainable above its cost of equity or just a strong quarter in a strong market.

Bank of America

Bank of America posted revenue of $31.6 billion, up 15% from a year ago, net income of $9.1 billion, up 27%, and EPS of $1.21, up 34%. Trading strength was broad-based, with FICC revenue of $3.5 billion, up 9%, and equities revenue of $3.6 billion, up 70%, while investment banking fees rose 50% to $2.1 billion. The bank bought back $6 billion of stock this quarter. BofA is the steadiest compounder of the group, and in a Damodaran-style equity model that steadiness is the point, a lower cost of equity because of that consistency can be worth more than a flashier growth number.

JPMorgan

JPMorgan posted revenue of $58.0 billion, up 27% from a year ago, net income of $21.2 billion, up 41%, and EPS of $7.70, up 47%, or $6.14 on an adjusted basis excluding a $4.6 billion gain tied to its Visa share conversion. Equity Markets revenue hit $6.0 billion, up 86%, and investment banking fees climbed 30% to their highest level since 2021, a point CEO Jamie Dimon highlighted directly. With ROE near 24% and ROTCE near 29%, JPMorgan remains the benchmark the rest of the industry gets valued against, the bank compounding furthest above its cost of equity, which under Damodaran's framework is exactly why it trades at the premium it does.

Takeaway: Every bank beat, every bank pointed to the same AI-fueled capital markets surge, and every CEO, in their own way, just told you not to expect an encore. When the people cashing the biggest checks of their careers are the ones warning you it can't last, that's worth remembering the next time this quarter gets used as the new baseline.

Overview

For two years, Nvidia was the only way to buy the AI boom without buying a data center. As of last Friday, July 10th, there's a second option, and it spent its first day trading under a ticker that looks like a typo.

SK Hynix, the South Korean memory chipmaker that quietly became one of the most important companies on the planet, listed on the Nasdaq Friday and raised $26.5 billion. That is the largest US debut ever by a foreign company. It beats Alibaba's $25 billion from 2014, a record that stood for twelve years and got erased by a company most retail investors couldn't have spelled a month ago.

For scale: this is the second-largest share sale in US history, full stop. The only thing bigger is SpaceX's $86 billion in June, which isn't really a comparison so much as a reminder of what kind of year this is.

The Deal

Hynix sold 177.9 million ADRs at $149 apiece. Demand came in at more than seven times the shares on offer, a number the bankers will call "strong demand" and the client will, correctly, call "you mispriced my company." The stock opened 14% above the offer and closed the day up about 13%. Last Friday, it traded under the placeholder ticker SKHYV and flipped to SKHY when regular trading opened on Monday. Somewhere on a 40th floor, a managing director is explaining to an analyst that the 13% pop was "the plan all along," which is true the way every roadshow is a triumph in hindsight.

Management

Hynix is run by CEO Kwak Noh-jung and controlled by South Korea's SK Group (chairman Chey Tae-won), the semiconductor arm of a conglomerate that has staked its future on AI memory. The defining call was early: committing to co-develop HBM with Nvidia years before rivals took the format seriously. HBM, or High Bandwidth Memory, is the specialized chip that sits next to the processor inside every Nvidia GPU, moving data at speeds standard memory cannot match. It is the chokepoint of the AI supply chain. That head start is why Hynix briefly overtook Samsung in overall DRAM revenue for the first time in Q1 2025 at roughly a 36% share. Samsung has since reclaimed the overall DRAM lead, but in HBM Hynix still commands roughly 58 to 62% of the market and is not being chased so much as lapped.

Strategy

The New York listing is the smaller of two strategic moves. In June, Hynix formalized a multiyear partnership with Nvidia covering memory for its next-generation Vera Rubin systems, with no financial terms disclosed. That ties Hynix to the roadmap of the one customer that matters. Capacity is the binding constraint, and Hynix is spending to fix it: a $14.6 billion next-generation HBM fab in Cheongju, and a $3.87 billion advanced-packaging plant in West Lafayette, Indiana, with mass production targeted for the second half of 2028. The US plant is the tell. Washington wants HBM built on American soil, and a Nasdaq listing paired with an Indiana fab is Hynix reading the room. Chips are policy now.

Industry

Memory is having its best cycle in a decade. HBM capacity for 2026 is effectively sold out, and diverting wafers toward it has starved the conventional market: DRAM contract prices rose 90 to 95% quarter-on-quarter in Q1 2026 and another 58 to 63% in Q2, the tightest supply since 2011. Hynix sits at the center of it with an estimated 50 to 60% of the HBM market and a widening lead, with HBM4 now shipping at roughly mid-$500 a stack, about double the prior generation. The one rival keeping pace on quality, Micron, lacks the fab capacity to press it; Samsung has struggled to qualify its parts at Nvidia. Second place in a two-horse race is still a real business. It's just a worse pitch.

Takeaway: SK Hynix didn't just pull off the biggest foreign listing in US history. It cemented itself as the second pillar of the AI trade. If Nvidia builds the brain, Hynix builds the memory it thinks with.

Stocks closed out their first losing week in three, with the S&P 500 down 1.6%, the Nasdaq off 2.9%, and the Dow down 0.9%. Semiconductor stocks led the slide; the Philadelphia Semiconductor Index fell more than 20% from its late-June record as investors grew nervous about the pace of AI infrastructure spending, a worry sharpened by a new AI model from Chinese startup Moonshot that claims to match OpenAI and Anthropic's leading systems.

Renewed US-Iran hostilities over the Strait of Hormuz sent oil prices to a one-month high, lifting energy stocks while pressuring airlines and cruise lines exposed to fuel costs. Netflix dropped sharply Friday after in-line earnings came with guidance that fell short of expectations.

Energy, real estate, and consumer staples were the week's strongest sectors. Technology and industrials lagged.

Top Gainers

Travelers ($TRV) +8.9%, Q2 earnings crushed estimates; an 88% EPS beat drove a wave of analyst price target hikes.

Chevron ($CVX) +6.2%, rode the Iran-driven oil price spike as energy names rallied broadly.

Exxon Mobil ($XOM) +6.1%, same story; higher crude prices are expected to add billions to Q2 upstream earnings.

Top Losers

IBM -26.0%, its worst week on record after the CEO disclosed a surprise Q2 revenue miss days ahead of the scheduled earnings date, tied to clients shifting spend toward hardware over IBM's software and infrastructure offerings.

Cisco Systems ($CSCO) -7.7%, profit-taking after a strong post-earnings rally, compounded by investor caution on enterprise infrastructure spending.

Netflix ($NFLX) -6.0%, in-line Q2 results weren't enough; guidance came in below Wall Street forecasts.

Economic Data

Outlook: Next Week

Also this Week: ~80 S&P 500 companies are scheduled to report. The season has run hot so far; of the roughly 40 companies that have already reported, 87% beat estimates, and blended earnings growth is tracking near 23% year over year, well above the 14% analysts expected heading in.

Hyperscalers (Amazon, Google, Meta, Microsoft, Oracle) spent the last decade printing free cash flow, but the chart shows cash flow rolling over hard, just as semiconductor suppliers building their AI infrastructure (Nvidia, Micron, Broadcom, Applied Materials) are inflecting vertically. It’s the clearest sign yet that this AI buildout is less a platform story and more a wealth transfer to the picks-and-shovels sellers. If the lines keep crossing, the real winners of this cycle won’t be the names everyone’s watching.

The yield curve prices risk and growth across time. Whether it's steepening or flattening, and which end is moving, tells you what phase of the cycle we're in. Recall that a bond’s price and yield are inversely related.

Bear flattener: short-term yields rise faster than long-term yields, the mark of an aggressive Fed hiking early in a downturn. The front end gets crushed, the long end can't keep up. Duration is low at the short end, so the price damage stays contained: painful for cash yields, not for a bond portfolio's mark-to-market.

Bull flattener: long-term yields fall faster than short-term yields. The classic flight-to-quality trade: investors pile into long bonds expecting rates to stay low, usually during a recession or a scare. The same duration that punishes long bonds when yields rise rewards them when yields fall. This is when owning duration pays off most.

Bear steepener: long-term yields rise faster than short-term yields. Shows up late in an expansion, when inflation runs hot, and the market starts pricing in tighter policy, or when investors demand more term premium, the extra yield they charge for the risk of holding duration over a longer, less certain stretch of time. This is where duration bites hardest: long bonds carry the most price sensitivity, so a bear steepener punishes long-bond holders exactly where the pain is amplified.

Bull steepener: short-term yields fall faster than long-term yields, usually because the Fed is cutting into a slowing economy. It marks the bottom of a recession, the point where easier money starts working through the system. Duration is low at the short end, so even a sharp drop in short yields barely moves bond prices. Cash and T-bills feel it in reinvestment rate, not account value.

Steepeners mark the start and end of a cycle. Flatteners mark the middle and the pivot. Watch which end moves; that's your duration risk. Short end moving is a rate story. Long end moving is a price story.

Quiz: Which yield curve regime is the U.S. government in currently?

*Answer at the end of this email*

Vacheron Constantin has been quietly outclassing its peers since 1755, the oldest watchmaker in continuous operation and the least interested in telling you so. It sits in the Holy Trinity with Patek Philippe and Audemars Piguet, but without the waitlist theater or the logo you'd clock across a room, just finishing precise enough to carry its own certification, the Geneva Seal.

The Overseas Chronograph is the house's answer to the Royal Oak and Nautilus, and the most understated of the three. A 42.5mm steel case carries the brand's Maltese cross worked into the bezel, a quiet signature rather than a shout. It runs on the in-house caliber 5200, a column-wheel automatic chronograph with a 52-hour power reserve, and the strap system is the real flex: a tool-free quick-release that swaps between steel bracelet, leather, and rubber in seconds, one watch built to move between a boardroom and a boat without missing a beat.

The only way to level up is to be in the arena, not planning around it. You can’t out-think your way past doing the work. The thinker who never acts is the worst person to be: all analysis, no motion. The doer who never thinks is better; at least they’re moving. An idiot in motion goes further than a genius at rest. But the rare, dangerous combination is both: a little analysis to point yourself in the right direction, then action, then small corrections as you go, never stopping to admire the map.

Some days, that path will be dimly lit, and you’ll have to walk it anyway. Every empire in history began with a first step on a day that looked completely ordinary. History rarely tells you when you’re living inside its first chapter. Act now.

Enjoyed this issue? Refer 3 friends, send us an email with the emails you referred, and unlock our DCF calculator!

How'd we do? We're still working out the kinks and will have more resources shortly to improve our research. We'll be posting updates on our X pages; give us a follow there too.

If you're interested in advertising, social management, or partnering on product or service affiliates, send an inquiry through the services page on our website, www.thestreetbrief.co