Newsletter · · Ashutosh Agarwal

Dealmaking Booms While Rising Yields Squeeze the Capital Markets Machine - The Capital-Markets Reopening - Week of August 4, 2026

A hawkish hold from new Fed Chair Kevin Warsh pushed long-term Treasury yields to a 19-year high and widened credit spreads, squeezing the debt-and-financing half of the capital-markets machine even as a rumored $400 billion AstraZeneca-Bristol Myers merger and a record deal pipeline kept advisory fees flowing, for the week of August 4, 2026.

The Capital-Markets Reopening

Week of August 4, 2026: Dealmaking Booms While Rising Yields Squeeze the Capital Markets Machine

Week of July 28 – August 4, 2026

TL;DR

  • The financing tap got harder to turn, and it happened on purpose. New Fed Chair Kevin Warsh held interest rates steady at his second meeting on July 29, but three of his own colleagues voted to raise, and Warsh refused to say what would make him move next. The bond market took that as a signal to sell: the 30-year Treasury yield jumped to about 5.2%, its highest level in roughly 19 years, corporate credit spreads (the extra yield companies must pay over the government to borrow) widened, and stocks slid. One well-followed strategist argued this was Warsh's plan all along, pull the Fed's thumb off the long end of the bond market and let borrowing costs rise to do the tightening for him. That is a direct headwind for the debt-and-financing half of the capital-markets machine.

  • And yet the deal pipeline just produced its biggest name of the year. The Financial Times reported that AstraZeneca is in early talks to buy Bristol Myers Squibb, a combination worth roughly $400 billion. Nothing is signed, and the antitrust hurdles are real, but a deal that size is a once-in-a-cycle advisory-fee event, and it lands on top of JPMorgan's own dealmakers telling a podcast that announced M&A is at "all-time highs." The demand for deals is not the problem right now. The plumbing around them is.

  • The exit door is now a warning light, not just a caution sign. SpaceX, the "textbook" mega-IPO the reopening thesis leaned on, has now fallen roughly 50% from its post-listing peak and trades below its $135 offer price, with a big share-lockup expiry looming that one investor called more important than earnings. Jersey Mike's got its billion-dollar IPO done but slipped on day one, and the Paramount–Warner Bros. deal was frozen by a court until as late as June 2027. The revenue engine still runs; the ability to cash out is where the stress is showing.

What's new

Ranked by what matters most for a trading book. Where the speaker runs the business they're describing (a CEO, CFO, or a bank's own dealmaker), it is flagged as an operator/insider; everything else is analyst, journalist, economist, or investor commentary. One honest note up front: none of our seven target firms put a senior executive on a podcast this week to talk about their own franchise, Goldman and Morgan Stanley already reported in mid-July, and the exchanges and boutiques report later. So the operator voices below come from other capital-markets firms (JPMorgan, Mizuho, CVC, CME), and our seven names surface mostly through deal mechanics and read-through. Those gaps are flagged plainly.

1. Warsh held rates, and deliberately let the long end of the bond market rip. This is the single most important development for capital markets this week, because it raises the cost of the debt that funds deals. On July 29 the Fed kept its policy rate at 3.5%–3.75%, but three regional Fed presidents (Beth Hammock, Lori Logan and Neil Kashkari) dissented in favor of a hike, the first three-dissent meeting since 2016, per Morning Brew Daily, "Investors Question Kevin Warsh's Credibility & People Are Talking Less?" (July 30). The market's reaction was violent for a "no change" meeting: on Bloomberg Surveillance, "Instant Reaction: Fed Chair Kevin Warsh's News Conference" (July 29), Apollo's Torsten Slok described a credibility problem as the 30-year Treasury yield jumped about 11 basis points to 5.20%. The best explanation of why came on Forward Guidance, "The AI Unwind And Warsh's Long-End Gamble | Weekly Roundup" (August 3): the hosts relayed Bank of America rate strategist Mark Cabana calling it a "classic central bank credibility shock", "the long end and stocks turned together roughly 3 PM while Warsh was still speaking", and then argued Warsh actually wants this. In their reading, Warsh signaled he'll shrink the Fed's balance sheet to "remove the Fed's footprint" from the Treasury market and "let the long end of the curve price to fair market value," which "restricts financing conditions" and widened credit spreads on cue. Why it moves the thesis: higher long-term rates and wider spreads are a tax on every leveraged buyout, every bond deal, and every refinancing. If Warsh keeps his thumb off the long end, the debt-funded part of the reopening gets pricier from here, by design.

2. A ~$400 billion drug mega-merger surfaced, the single biggest advisory-fee event of the year, if it happens. On Squawk on the Street, "9am Hour: U.S.-Japan Yen Intervention... AstraZeneca-Bristol Myers M&A Buzz 8/3/26" (August 3), CNBC's David Faber reported that "the FT, Financial Times, reporting AstraZeneca in talks to buy Bristol-Myers... both sides are said to have held discussions in recent months," and that the potential cost savings "are certainly in the many billions." Add the two companies' market values together and you get around $400 billion. On the follow-up hour, Squawk on the Street, "10AM Hour: Brookfield CEO on Data Centers... Potential Pharma Mega-Deal 8/3/26" (August 3), CNBC's Annika Kim Constantino laid out the catch: regulators would scrutinize the overlap in cancer drugs, the two firms have competing lung-cancer treatments and Bristol's immunotherapy Opdivo, and Bristol is staring at big patent expirations (its blood-thinner Eliquis). On The Rundown, "The US Bails Out the Yen, AstraZeneca Explores a Bristol Myers Mega-Merger" (August 3), the hosts warned antitrust review "could tie up the deal for years." Why it matters: a $400 billion combination is the kind of deal that pays boutique and bulge-bracket advisers alike for years and pulls through financing fees on top. It is exactly the pipeline the reopening thesis is built on, but the antitrust risk is a reminder that the very biggest deals are also the hardest to close. Treat it as a rumor with real fee optionality, not a done deal.

3. JPMorgan's own dealmakers say announced M&A is at all-time highs, with a "shadow backlog" behind it. This is the closest thing this week to operator-grade proof that the deal engine is real. On Making Sense, "Deals and discipline: What's driving markets at mid-year?" (July 30), JPMorgan's Charlie Bukart, the bank's global head of advisory and M&A (operator/insider), said "activity level is very high. We're at all-time highs in terms of announced M&A volumes," and that "when you look at pipelines across the street, including ours, as well as the shadow backlogs that people have, it really feels like activity levels are going to remain very elevated into the rest of the year." (A "shadow backlog" is the pile of deals being worked on privately that hasn't been announced yet, a forward tell on future fees.) He added that cross-border M&A is "at multi-decade highs," driven by foreign buyers wanting more exposure to North America. But he threw cold water on the easy narrative that regulators have simply rolled over: "the perception has probably run ahead of the reality" on a friendlier Washington, though "when we look at approvals for deals, they are quicker and the number of investigations is down." His colleague Evan Junick added a useful behavioral data point: among "the 30 most AI-exposed companies," references to return on invested capital rose "about five times" in the first quarter, a sign management teams are starting to be asked to justify the AI spending spree. Why it matters: this is the demand side of the fee pool talking, and it's saying the same thing the record deal calendar is: volumes are high and the backlog is deep. The nuance, approvals faster, but the "friendly regulator" story oversold, is the exact tension the AstraZeneca–Bristol deal will test.

4. An AI-driven market wobble hit a $45 billion hedge fund, and Goldman's prime-brokerage desk sat in the middle of it. The week's other big story was a sharp AI-stock unwind, and it produced a spectacular blow-up. On Squawk on the Street, "10AM Hour: Leopold Aschenbrenner's Hedge Fund Situational Awareness Sells All Public Positions, Jersey Mike's CEO on IPO... 7/30/26" (July 30), David Faber reported that Leopold Aschenbrenner's AI-focused fund, Situational Awareness, had a net asset value of about "$45 billion" as of July 1, "a fund that started the year at $3 billion", and was selling all its public positions. On Squawk on the Street (August 3), Faber noted the fund "was amongst, if not the leading fee payer in the prime brokerage for Goldman." (Prime brokerage is the business of lending to and clearing trades for hedge funds, a steady, high-margin revenue line for the big banks, and one that swells when funds use a lot of borrowed money.) Why it matters: it cuts both ways for Goldman. A marquee client unwinding is a hit to a lucrative fee stream and a reminder that prime-brokerage revenue is only as safe as your clients' leverage. But an orderly liquidation of a $45 billion book with no contagion headlines is also a quiet advertisement for the desk's risk management. Watch whether the AI unwind stays contained or starts to pull down the trading and financing revenue that carried the banks' blowout second quarter.

5. The bond market is being asked to swallow a record wall of AI debt, right as rates back up. On Beyond The Obvious, "Markets Mindset: Q3 2026 Debt Capital Markets: Absorbing the Tech Issuance Surge" (July 29), Mizuho's Victor Forte and Moshe Tomquitz (operators who run investment-grade and debt capital markets) put hard numbers on it: year-to-date high-grade bond issuance is "1.3 trillion, up... 34% on an absolute basis," with technology issuance "up 117%," utilities "up 47%," and communications "up 277%" as companies borrow to build AI infrastructure. They counted roughly "$54 billion" of data-center-backed structured bonds since last year, with total supply potentially reaching "$80–90 billion." Crucially, they said "August has potential to be a record month" as issuers rush to sell bonds before the September "blackout" (the quiet period around the Fed meeting), and that "spreads have hit a pretty hard floor" unless the market decides 2027 is the peak of AI spending. On Know More. Risk Better., "From Know More to Above the Fray: Markets & the Road Ahead" (July 30), the hosts noted almost "$200 billion" of hyperscaler debt already issued against a projected "$825 billion" of capex this year, and flagged a high-yield data-center bond that had to be priced at a steep discount as an early sign of buyer fatigue. Why it matters: debt underwriting is a core fee line for the banks, and a record August calendar is good news for near-term revenue. But "spreads at a hard floor" plus a rising long-end yield is the recipe for the financing window to get choosier, the same warning the equity side is flashing.

The debate

The core question hasn't changed, but the balance of evidence shifted this week. Is this a durable, multi-year reopening of capital markets, or a fragile head-fake that cracks the moment the AI trade or the rate backdrop turns?

The durable case. The demand for deals is not in doubt. JPMorgan's own head of M&A says announced volumes are at all-time highs with a deep shadow backlog and cross-border activity at multi-decade highs (Making Sense, July 30). The year produced its biggest potential deal yet in the rumored AstraZeneca–Bristol Myers combination (Squawk on the Street, August 3). Private equity is flush and, more importantly, returning cash: on Dry Powder: The Private Equity Podcast, "Looking Over the Horizon w/ CVC's Rob Lucas" (July 29), CVC CEO Rob Lucas (operator) said the firm raised "over 23 billion euros over the last 12 months" and returned "over 50 billion euros of capital... at four times multiple of money in the 29% IRR" over five years, and that fundraising ability is "very, very closely linked to the ability to make realizations and return capital." Every one of those realizations is an IPO or a sale, and every one is a fee. And the bond calendar is on track for a record August (Beyond The Obvious, July 29). Even the newest twist, the return of the SPAC, or blank-check company, points to a wider set of exit routes: on Accounting Matters, "Return of the SPAC: What's Different This Time" (August 4), the hosts noted "138 SPACs raising $26 billion in 2025 representing 40% of U.S. IPO deal count," now bringing more mature, PE-backed companies to market.

The fragile case. The plumbing around all this demand deteriorated visibly this week. Rates: Warsh's hawkish hold pushed the 30-year yield to a ~19-year high near 5.2% and widened credit spreads, and the September rate-hike odds jumped, on RiskReversal Pod, "Warsh Out in Bonds + CME Group's Tim McCourt on Single Stock Futures" (July 31), Dan Nathan cited the CME's own tool pricing "a little more than a 70% probability" of a hike at the September meeting. Exits: on Bloomberg Surveillance, "Bloomberg Surveillance TV: August 3rd, 2026" (August 3), Clio Capital's Sarah Kunz noted SpaceX is "down over 30% since its IPO" and faces a staggered share-lockup expiry she called "more important than earnings", and warned it "may color hesitation around potential OpenAI and Anthropic IPOs." The IPO surge itself drew skepticism from Goldman's own podcast: on Exchanges, "Is the Surge in US IPOs a Warning Sign for Investors?" (July 28), the University of Florida's Jay Ritter argued "the pickup in US IPO activity is more of a return to normal than a true IPO boom", record dollars raised, but only "a little over 100 companies per year" versus "over 300 companies per year in the 80s and 90s", while Acadian's Owen Lamont warned that a genuine wave of issuance "would be a symptom to me that the market is overvalued and we're in a bubble," because "firms want to sell equity when equity is overpriced." And credit is starting to twitch: on Saxo Market Call, "Wishy washy Fed Chair Warsh vibe punches sentiment in the gut" (July 30), high-yield spreads widened 7 basis points to 289, still historically tight, but no longer falling.

My read. Last week the story was that the revenue engine roared; this week the story is that the engine still runs but the road got rougher. The demand for deals, strategic M&A, sponsor exits, a record bond calendar, is genuinely strong, and the AstraZeneca–Bristol rumor shows the biggest fish are still swimming. But the two things that convert demand into completed, cashed-out fees, a workable exit window and an affordable cost of debt, both got worse this week, and the Fed did the second one on purpose. That argues for the same posture as last week, only more so: favor the elite strategic-advisory franchises (which get paid to advise regardless of the financing weather) over the volume leveraged-finance and pure-IPO-window plays, which sit directly in front of the rate and exit stress. The single biggest swing factor from here is whether Warsh keeps letting the long end run. If he does, this becomes a story about deals that get announced but take longer, and cost more, to close.

Stocks in play

Reminder: our seven names got no dedicated franchise coverage on the podcasts this week (Goldman and Morgan Stanley reported in mid-July; ICE, Nasdaq, Evercore, Moelis and Jefferies report later). What follows blends the read-through from this week's podcasts with each name's setup. Genuine coverage gaps are flagged, not papered over.

Goldman Sachs (GS). Bull: Goldman touches almost every strong current, it was a lead underwriter on the year's marquee IPOs (SpaceX), a top-tier adviser positioned for a deal wave like AstraZeneca–Bristol, and runs a huge prime-brokerage and trading book. Its asset-management arm is also scaling in private markets: on Alt Goes Mainstream, "Goldman Sachs' Kyle Kniffen - the arc of alternatives at Goldman Sachs" (July 30), Goldman's global head of alternatives for third-party wealth (operator) said Goldman Sachs Asset Management is "a top five asset manager on the public side and a top 10 asset manager on the private side of $625 billion," a pioneer in GP Stakes (buying minority stakes in other fund managers) with a new T. Rowe Price partnership rolling out later this year. And the stock looks washed out to some: on RiskReversal Pod (July 31), Guy Adami said Goldman is "for sure overdone to the downside," noting it made an all-time high around "1153" about two weeks earlier and had since fallen "below $1,000." Bear: the same breadth is a liability when the plumbing tightens, the SpaceX collapse dents underwriting momentum, the Situational Awareness unwind removes a top prime-brokerage fee payer, and higher long-end rates cool the leveraged deals it finances. Catalyst / number to watch: whether the AI unwind stays contained (so trading and prime revenue hold up), and whether the advisory backlog converts into announced deals in Q3.

Morgan Stanley (MS). Bull: like Goldman, Morgan Stanley co-led the SpaceX IPO and is a natural winner in a strategic-advisory and equity-underwriting upcycle, with a giant wealth-management arm that captures the wealth these deals create. The scale of the fees is concrete: on The Powers That Be: Daily, "Zaz in Limbo" (August 4), author Bill Cohan said SpaceX's IPO "raised $86 billion in equity, valued the company at $1.8 trillion," and "23 separate underwriters split $555 million in fees. Goldman, Morgan Stanley each got $111 million." Bear: SpaceX's collapse from about "$225 a share after four days" to roughly half that (per Cohan) is a bad look for the deal's leads and a live example of the after-market turning hostile, and Morgan Stanley's wealth inflows are partly a one-off function of IPO wealth that slows when the window narrows. Catalyst / number to watch: the SpaceX lockup expiry (a wave of newly sellable shares that could pressure the stock further), and whether the OpenAI/Anthropic IPO calendar survives the SpaceX experience.

Intercontinental Exchange (ICE). Bull: the read-through is favorable, a record wall of AI-related bond issuance (Beyond The Obvious, July 29) plus a volatile rate backdrop feeds fixed-income data, futures and trading volumes, and ICE owns the New York Stock Exchange, which is pushing into new market structure. On Unchained, "Should Tokenized Stock Only Come From Issuers? Yes, Says Carlos Domingo" (August 4), Securitize's Carlos Domingo said his firm is the approved transfer agent and design partner for "the New York Stock Exchange's digital ATS launching in Q4" with "tens of different instruments", a new listings-and-trading venue for tokenized assets. Bear: higher-for-longer rates keep the mortgage-data segment tied to a frozen housing market, and a genuine IPO slowdown would eventually crimp listings. Catalyst / number to watch: Q2 results (not yet reported) and the transaction-versus-recurring-data revenue mix. (No ICE-specific earnings coverage this week, an honest gap; the only ICE-adjacent item was the NYSE digital-venue mention above.)

Nasdaq (NDAQ). Bull: the listings franchise still has a pipeline behind it, Jersey Mike's just priced, and Anthropic and OpenAI are the marquee names still expected. On The Best One Yet, "'AI Can't Bake Bread', Jersey Mike's IPO..." (July 31), the hosts noted Jersey Mike's raised "$1 billion with shares 10 times oversubscribed", the second-largest restaurant IPO in history, evidence investors still want tangible, cash-generating businesses. Bear: listings revenue is reflexive, and the after-market has turned hostile, Jersey Mike's fell "6% on its first trading day" at a "$7.3 billion valuation" (Morning Brew Daily, "...Jersey Mike's Goes Public" (July 31)), and SpaceX's collapse is exactly the kind of signal that makes the next issuer hesitate. Catalyst / number to watch: whether Anthropic and OpenAI still come to market on schedule, and the Q2 listings-and-data revenue mix. (No Nasdaq-specific coverage this week, the franchise surfaced only through the IPO pipeline.)

Evercore (EVR). Bull: a pure strategic-advisory shop is the cleanest way to play a deal wave that doesn't depend on cheap financing, and this week handed it exactly that, the rumored AstraZeneca–Bristol Myers combination plus JPMorgan's "all-time high" announced-M&A commentary (Making Sense, July 30) is pure boutique-advisory fuel. Bear: boutiques earn on completed deals and have no trading or lending cushion, so a rate-driven slowdown in closings, or a mega-deal that stalls in antitrust review, shows up in their results faster than at the diversified banks. Catalyst / number to watch: Q2 advisory backlog and managing-director productivity. (No Evercore franchise coverage this week, an honest gap.)

Moelis (MC). Bull: a pure-play adviser that captures the same rich M&A pool, with a valuable restructuring option if the AI-linked debt and private-credit leverage building across the system ever turns to stress, restructuring advisory tends to boom exactly when financing breaks. Bear: with no trading or lending ballast at all, Moelis is the most exposed of the seven to a deal-volume air pocket. Catalyst / number to watch: Q2 advisory revenue and any shift toward restructuring in the mix. (No Moelis-specific coverage this week, an honest gap.)

Jefferies (JEF). Bull: leveraged finance plus advisory gives Jefferies real leverage to a genuine reopening, and its off-calendar fiscal quarter often makes it the group's earliest read on where fees are heading. Bear: it sits most directly in the path of any leveraged-finance slowdown, and this week's combination of a rising long-end yield, widening spreads and a record-but-fatiguing bond calendar is precisely the backdrop that pressures that business first. Catalyst / number to watch: its next print as the group's leading indicator. (No Jefferies-franchise coverage this week; the only mention was Jefferies analyst Brent Thill commenting on Microsoft's earnings on Squawk on the Street (July 30), sell-side research, not the firm's own business.)

Read-throughs

Boutique advisers (EVR / MC / JEF). The strategic-deal environment is unambiguously busy at the top end, the rumored AstraZeneca–Bristol Myers combination (~$400 billion) sits on top of JPMorgan's "all-time high" announced-M&A commentary and a cross-border wave at "multi-decade highs" (Making Sense, July 30; Squawk on the Street, August 3). The nuance that matters for the boutiques: JPMorgan's Bukart warned the "friendly regulator" perception "has probably run ahead of the reality," even though approvals are quicker and investigations fewer, and the AstraZeneca–Bristol antitrust overhang is the live test of which is true. Boutiques win the advice fees regardless of financing costs, which is why they screen better than the leveraged-finance names in this particular backdrop; the risk is that a mega-deal stalling in review pushes fee recognition out by quarters.

Leveraged finance & private credit. This is where the week's stress concentrates. The AI-debt build is enormous and beginning to strain: Mizuho's desk counted year-to-date high-grade issuance up 34% in absolute terms with tech up 117%, and up to $80–90 billion of data-center bonds, but said "spreads have hit a pretty hard floor" (Beyond The Obvious, July 29); a high-yield data-center bond had to price at a discount (Know More. Risk Better., July 30); and high-yield spreads have stopped falling (Saxo Market Call, July 30). Layer Warsh's deliberate push to lift the long end on top, and the cost of leveraged debt is rising from both ends. Near-term this still means fat underwriting fees, August could be a record issuance month as borrowers rush ahead of the September Fed blackout, but it's a "get it done now" scramble, not a healthy, open-ended window.

PE sponsors monetizing exits. The exit math is the pinch point. Sponsors are cash-rich and under pressure to return capital, CVC's Rob Lucas tied fundraising directly to "the ability to make realizations and return capital" and flagged industry "consolidation... within the GP community" as capital gravitates to the biggest, most-trusted managers (Dry Powder, July 29). But the two main exit routes both wobbled: the IPO after-market turned hostile (SpaceX down ~30%+, Jersey Mike's off on day one), and a court froze one of the year's biggest strategic sales. Watch the workarounds, the SPAC revival (Accounting Matters, August 4) and continuation vehicles are sponsors finding side doors when the front door sticks. Sponsors will still monetize; the question is at what discount, and how much of the proceeds get recycled straight back into new deals.

Exchange listings & data. For Nasdaq and ICE, the twin engines are listings and data. Listings are reflexive and currently fragile, the after-market is punishing new supply, which is a near-term drag on the listings line. But the data-and-trading side is the durable winner from this week's chaos: a record bond calendar, a violent rate move, and an AI-stock unwind all mean more volume across fixed-income data, futures and equities. New market structure is also expanding the pie, CME is building out single-stock futures (on RiskReversal Pod (July 31), CME's Tim McCourt (operator) described "55 names" in single-stock futures and "22 names" in micro contracts, with "over 130 retail partners") and NYSE is launching a digital tokenized-asset venue in Q4 (Unchained, August 4). The exchanges are the closest thing to a "sell the volatility, not the direction" trade in this group.

What changed vs. last week

Last week's issue (July 14–21) was a victory lap: five of the six largest US banks posted a blowout quarter, revenue up 39%, Goldman's equity-trading up ~50% past estimates, and, for the first time this cycle, CEOs and CFOs put their names on the reopening while warning it was "as good as it gets." The two soft spots then were narrow: SpaceX had just dipped below its offer price, and a judge had frozen the Paramount deal for 14 days. This week, the plumbing problems the operators warned about started to bite. Five updates:

  • The rate backdrop flipped from a mild tailwind to an on-purpose headwind. Last week a cool June inflation print had knocked near-term hike odds down and the story was "no urgency to move." This week Warsh held but let the long end rip, the 30-year yield hit a ~19-year high near 5.2%, credit spreads widened, and one strategist argued Warsh is deliberately using balance-sheet tightening to raise borrowing costs (Forward Guidance, August 3). September-hike odds jumped past 70% on the CME's own gauge (RiskReversal Pod, July 31). The financing half of the fee pool now faces a rising, not a stable, cost of money.

  • SpaceX went from "just below offer" to a full ~50% round-trip. Last week SpaceX had merely slipped under its $135 offer price. This week it's down roughly 30%-plus from its IPO and about half its ~$225 post-listing peak, with a lockup expiry looming that an investor called "more important than earnings" (Bloomberg Surveillance, August 3; The Powers That Be, August 4). The exit window is no longer just tight, it's actively burning the retail buyers who chased the marquee listing.

  • A brand-new mega-deal appeared, the deal engine kept refilling even as financing tightened. Last week's deal news was a record share of $10 billion-plus transactions. This week produced the single biggest name of the year: AstraZeneca in reported talks for Bristol Myers, roughly $400 billion combined (Squawk on the Street, August 3). Demand for deals didn't fade with the rate move, which is the crux of this week's split-screen.

  • The Paramount freeze hardened into a year-long delay. Last week a judge had paused the $110 billion-plus Paramount–Warner Bros. deal for 14 days. This week the timeline blew out: the deal is now delayed until as late as June 2027, with mounting "ticking fees", on The Powers That Be (August 4), Bill Cohan explained the delay adds "75 cents a share... bringing the deal's total value to $31.75 per share, something like $113 billion," while the stock trades at a "15, 17 percent discount" as arbitrageurs doubt it closes (Organized Money, July 28). Even DOJ-cleared mega-deals can get stuck for a year, a real risk for advisory close rates on the biggest transactions.

  • A new stress line opened: an AI-hedge-fund blow-up hitting bank prime brokerage. This wasn't in last week's picture at all. Situational Awareness, a $45 billion AI fund up from $3 billion at the start of the year, liquidated its public book, and it was among Goldman's top prime-brokerage fee payers (Squawk on the Street, July 30 and August 3). It's a reminder that the trading and financing revenue behind last quarter's blowout is tied to client leverage that can unwind fast.