Newsletter · 2026-09-22 · Ashutosh Agarwal

Biggest Deal Year Ever as Anthropic Pauses Its $2 Trillion IPO - Capital Markets: IPOs, M&A & Exchanges - Week of September 22, 2026

Capital Markets: IPOs, M&A & Exchanges for the week of September 22, 2026. Podcast synthesis on record equity issuance and IPO proceeds, Anthropic delaying its $2 trillion listing, Goldman's Gene Sykes calling 2026 the biggest M&A year in history, the Paramount and Warner Bros. Discovery antitrust fight, Kalshi's push into perpetuals, and cracks in private credit.

Capital Markets: IPOs, M&A & Exchanges

Week of September 22, 2026: Biggest Deal Year Ever as Anthropic Pauses Its $2 Trillion IPO

The busiest dealmaking year ever, and the $2 trillion IPO that hit the brakes.

Week of September 22, 2026. This week's podcasts served up a paradox: bankers are calling 2026 the biggest year for deals and new listings in living memory, even as the single largest IPO on the runway just delayed itself, private-credit defaults hit a record, and the bond market started pushing back. Below: what the people who actually run these markets said, in their own words, and why it matters.

Big Story 1: A Record-Smashing IPO Year, With the Biggest Deal of All Now on Pause

Start with the scoreboard, because it is genuinely historic. On JPMorgan's own podcast, the bank's global head of tech equity capital markets (the business of helping companies sell shares to the public), Eddie Bayoun, walked through numbers that would have sounded made up a year ago.

• Global equity issuance is running near $800 billion so far in 2026, up 75% from a year ago, with just over half of it in the US. US equity issuance alone is $430 billion, more than double the roughly $200 billion at the same point last year, and already on track to blow past 2021's full-year record of $458 billion (J.P. Morgan Making Sense, "What's next for tech equity capital markets in a record year for mega-cap IPOs?", September 4, 2026).
• IPO proceeds have already reached $162 billion, more than the $152 billion raised in all of 2021. Bayoun expects the full-year figure to clear $200 billion.

To grasp how much the plumbing has changed, Bayoun reached back to the old record. "The largest US IPO ever in the market was Alibaba back in 2014, when we raised 25 billion," he said. "We literally had to scour the earth, mapping investor by investor, portfolio manager by portfolio manager, analyst by analyst, to really get to that 25 billion." This year's mega-deals (SpaceX, the biggest IPO ever; SK Hynix, the biggest-ever listing by a non-US company; plus giant follow-on sales from Alphabet and Intel) were "even far larger than what Alibaba had done," and the demand showed up "often in the first day alone." JPMorgan was one of only two banks on all four.

His colleague Jack Atherton made the point that some of these are simply too big to ignore: SpaceX, he said, is "very clearly an end-of-one company... every investor needs to have a view, as difficult as it is to think about valuation for a company that is trying to colonize Mars." (J.P. Morgan Making Sense, September 4, 2026)

The twist: the largest IPO still to come just delayed itself. Anthropic, the AI lab, had been lining up what could be the biggest listing in history, a target valuation around $2 trillion, roughly double its ~$965 billion valuation in May, with marketing pegged for mid-October. Then it pulled back.

• On the Elon Musk Podcast, "Anthropic delays $2 trillion IPO for safety" (September 20, 2026), the hosts laid out the oddity: Anthropic is "sitting on $100 billion in annualized revenue, a threshold that dictates an immediate public market debut in almost every scenario in financial history," and is "deliberately pushing the timeline back." The stated reason is AI safety: the difficulty of proving to pension and mutual funds that an increasingly autonomous system "will not go rogue."
• The fix is unusual: a partnership with Accenture, with each side committing at least $1 billion, to embed independent evaluators (a group called Faculty) inside the model-building process to "red team" it before launch. The hosts flagged the obvious problem: "You have the entity seeking approval paying the entity granting the approval. The independence of an evaluator is fundamentally compromised the moment their funding relies on the success of the company they are evaluating."
• Anthropic's revenue ramp is the reason the number is so large, and so debated. Reports put annualized revenue north of $100 billion, up from single-digit billions barely a year earlier. On the Rich Habits Podcast, "Trump's $5K Bribe, Anthropic's $2T IPO, & Meta's Muse" (September 11, 2026), the hosts noted the same company carries an estimated $10–15 billion in cumulative losses through 2025 and declining gross margins even as revenue explodes.

Anthropic is not alone in stalling. On Squawk on the Street, "AI Leaders Call For Slowdown... Altman Rules Out 2026 IPO" (September 14, 2026), OpenAI's Sam Altman was reported to have ruled out an OpenAI IPO in 2026, floating 2027 instead; CNBC's David Faber said Anthropic still looked likelier to go public first. Goldman Sachs and Morgan Stanley shares fell 2%+ that day on worries that delayed mega-IPOs would dent banking fees.

Why it matters: The public market has never been able to absorb this much equity this fast, but the pipeline is now top-heavy with a handful of AI names whose economics are still being argued over. If the biggest one keeps slipping, the "record year" starts leaning on a promise rather than a print.

The rest of the IPO world, briefly: China's Unitree Robotics raised about $904 million on Shanghai's tech board and closed up 460% on debut for a ~$50 billion value; India's National Stock Exchange is marketing what would be the country's largest-ever IPO at a ~$55 billion valuation (Market Maker, "China's Robot IPO Boom & India's Biggest IPO Ever Explained," August 24, 2026). Not everything worked: fast-fashion group Shein limped to a Hong Kong listing around $26–27 billion, down from $100 billion-plus in private markets in 2022 (The Rundown, "Amazon Sued Over Ad Prices, Shein's IPO Falls Flat," September 1, 2026). Smart-ring maker Oura is targeting a September listing north of $16 billion with Goldman and Morgan Stanley leading (Rich Habits Podcast, "Nvidia's $675B Guidance, Oura Ring IPO & Meta's Lawsuit," August 28, 2026).

Big Story 2: The Biggest M&A Year in the History of the World, and the Regulators Standing in the Way

If one quote sums up the mood among dealmakers, it came from Gene Sykes, Goldman Sachs' co-head of global M&A, sitting for a rare long interview.

"This is the biggest M&A year in the history of the world. Biggest year by far. We expect the M&A market to be much bigger than it was at the peak. The last peak was 2021, inflated by COVID." (Gene Sykes, on Odd Lots, "A Goldman M&A Banker Helped Bring the Olympics to Los Angeles," September 19, 2026)

Sykes made two points worth holding onto:

• The buyers have changed. For years, roughly 40% of M&A was private-equity firms selling companies they owned. This year that's down to 30%, because "there's so much more substantive demand": companies across semiconductors, power generation and natural resources buying to build out the infrastructure for AI. In a strange inversion, he noted, the private companies (OpenAI, Anthropic, SpaceX) now feel like the incumbents, and they're "almost too big for anybody to buy."
• Most startups are built to be sold. "For 80% of the startups, the end game is selling themselves to somebody else," Sykes said. The other 20% survive as independent companies only if they have something "totally differentiated." He also threw cold water on the AI-productivity hype: by Goldman's own analysis, just 2% of companies have told investors AI is actually helping earnings per share, though 20–30% have started implementing it seriously.

Sykes' warning to anyone assuming the party lasts: boards are "pro-deal in a way that we see from time to time. But I don't think we see it as something that lasts forever... You have to take advantage of it when it's there."

The hard numbers back the mood. On Market Maker, "Investment Banking Interview Prep: The M&A Deals You Need to Know in 2026" (September 14, 2026): the first half of 2026 saw about $3 trillion in deal value, with Q2 hitting a record $1.6 trillion; there were 47 mega-deals above $10 billion worth $1.3 trillion combined. Named blockbusters include SpaceX–XAI ($250B), the Unilever Foods–McCormick reverse-Morris-Trust ($44.8B), NextEra–Dominion ($67B) and Shell–ARC Resources ($22B). Goldman topped the league tables with roughly $1.4 trillion of deals and about $3 billion in fees. Corporate CEOs are leaning in: Eli Lilly's Dave Ricks said the drugmaker has announced "over $25 billion" in deals year-to-date, more than all of last year (Squawk on the Street, "Exclusive Interviews with the CEOs of Goldman Sachs and Eli Lilly," August 31, 2026), and Morgan Stanley's M&A co-head John Collins said healthcare deal volume has already matched all of 2025 in eight-and-a-half months, without any true mega-deals (Squawk on the Street, "Bessent Testifies, Thermo Fisher CEO, Morgan Stanley's M&A Outlook," September 15, 2026).

The wall in the way is antitrust, and the fight over Paramount's $110 billion purchase of Warner Bros. Discovery has turned into the year's marquee regulatory brawl. Even after federal approval, California Attorney General Rob Bonta is leading a coalition of a dozen states suing to block it, arguing behavioral promises (Paramount's pledge to release 30 films a year) aren't enough and demanding a structural break-up, potentially spinning off a cable channel like CNN (Strictly Business, "Who is Rob Bonta and Why is California's AG Doggedly Pursuing Paramount?," September 2, 2026). Trial is set for early March 2027.

Then it escalated. On the Elon Musk Podcast, "Paramount's twenty-one billion dollar California exit" (September 17, 2026), the hosts detailed Paramount CEO David Ellison's threat to pull the studio out of California entirely, a move a leaked LA County report estimates would cost the state $21 billion a year and nearly 58,000 jobs, from carpenters and electricians to caterers and lumber yards. And the Justice Department has waded in on Paramount's side, asking a federal judge to force the states to post a financial bond to cover the deal's delay costs, the DOJ's logic being that state enforcers acting as "rogue veto points" on a $110 billion interstate deal should have to "put their own money on the line." A two-day mediation with Bonta and the Writers Guild is scheduled.

Why it matters: The deal engine is running hot for a genuinely new reason: AI is reordering entire industries and forcing incumbents to buy their way in. But the biggest media consolidation in a generation is now a live test of whether individual states can veto a nationally-approved merger, and the answer will shape how every future mega-deal gets structured.

Antitrust footnote: Not every enforcer is winning. A federal judge found Google holds an illegal monopoly in ad technology but refused the DOJ's demand to force a sale of its ad exchange, opting for behavioral fixes instead, the third straight courtroom setback for antitrust enforcers against big tech (Elon Musk Podcast, "Google keeps its illegal ad tech monopoly," September 3, 2026). And in a cautionary tale for anyone trying to duck review, the FTC hit Edwards Lifesciences and Genesis MedTech with a combined $12 million penalty, the largest failure-to-file fine on record, for structuring a $141.8 million acquisition to slip below the reporting threshold (Our Curious Amalgam, "Do I Really Need to File? HSR Avoidance and the Edwards/Genesis Settlement," August 24, 2026).

Big Story 3: Prediction Markets Grow Up, With Perpetuals, the Supreme Court, and a Robinhood Milestone

Prediction markets (platforms where you can buy contracts that pay out based on whether some future event happens) spent the last month looking less like novelty sportsbooks and more like real exchanges. Kalshi is leading the charge into products that would make a traditional futures exchange nervous.

On Bloomberg Businessweek, "Kalshi Launches Gold & Silver Perpetuals" (September 14, 2026), Kalshi's chief risk officer Udesh Jha (who spent 16 years at exchange giant CME Group) laid out the expansion:

• Kalshi's crypto "perpetuals" (futures contracts with no expiration date, letting traders use leverage around the clock) have already done about $45 billion in cumulative trades. It just launched 24/7 gold and silver perpetuals, an asset class Jha called "15 to 20 times the size of crypto."
• Next: Kalshi is "in the process of filing in short order" to offer the country's first regulated perpetual futures on individual stocks, around 60 contracts tied to names like Tesla, Apple and Nvidia (companies worth at least $100 billion).
• The pitch is regulation as a moat. "We have seen some of these equities as well as metals and other perpetuals grow significantly in offshore unregulated channels," Jha said, positioning Kalshi (overseen by the CFTC) against offshore venues like Hyperliquid, where "trades can get auto-liquidated" with "very limited risk protection."

But the legal ground is shifting. On the Elon Musk Podcast, "Kalshi ruling triggers Supreme Court showdown" (August 30, 2026), the hosts explained a 9th Circuit court ruled 3-0 that Kalshi's sports contracts are "sports bets," not financial swaps, stripping the federal protection Kalshi had leaned on and creating a direct conflict with the 3rd Circuit's opposite ruling. That kind of "circuit split" is exactly what the Supreme Court exists to resolve; traders on rival platform Polymarket put the odds of the court taking the case at 55%, up from about 29%.

Kalshi is also learning what running an exchange means when someone games it. On Bloomberg Talks, "Kalshi's Brian Quintenz Talks George Santos Lifetime Ban" (August 31, 2026), board member and former CFTC commissioner Brian Quintenz described Kalshi's first-ever lifetime ban plus a $71,000 fine on former Congressman George Santos for manipulating contracts on State of the Union attendance. Quintenz also cited a striking data point: users aged 18–21 have traded $5.4 billion on Kalshi this year.

And the retail giant is now getting real money from all this. On Tokenized, "Visa Stablecoin Settlement Hits $20BN, Up 15X" (September 14, 2026), the hosts noted Robinhood generated $156 million in prediction-market revenue in Q2 2026, up roughly 10x year-over-year across 13.6 billion contracts, the first quarter in which prediction markets out-earned crypto trading on the platform.

Why it matters: Prediction markets are morphing into leveraged, always-on exchanges competing directly with CME and CBOE, a genuinely new venue category, while the courts still can't agree whether they're finance or gambling. A Supreme Court ruling would set the rules of the road for a business scaling into the tens of billions.

Quick Hits: What Else Moved the Plumbing of the Markets

Private-credit cracks are showing. Fitch's trailing-12-month private-credit default rate across 1,300 US borrowers hit a record 6.3% in August, with 45% of defaults coming from stressed maturity extensions. The stress is worst at the bottom: among borrowers with $10–20 million in earnings, 12% of loans now trade below 90 cents on the dollar, versus about 1% in 2023 (Eurodollar University, "BREAKING: Private Credit Defaults Just Hit a Record High," September 18, 2026). Blackstone is even running a secondary sale for investors in an ~$11 billion real-estate fund amid redemption pressure. The episode's line to remember: "Liquidity is not simply disappearing. It is being rationed."

The bond market is pushing back. FT columnist Katie Martin put it bluntly: "Governments are just borrowing too much damn money... the US national debt has, of course, crossed through $40 trillion for the first time." The 10-year Treasury yield topped 4.8% and the 30-year cleared 5.25%. Her sharper structural point: cash-hungry AI hyperscalers (Meta, Microsoft, Amazon) are now issuing so much corporate debt that they're "bossing around" smaller borrowers, and Anthropic and OpenAI are reportedly talking to rating agencies about investment-grade ratings: "even if you think you're not a tech investor, you're a tech investor" (Prof G Markets, "Why The Bond Market Is Starting To Revolt," September 11, 2026).

AI is now the main event in debt markets, too. At JPMorgan's European Leveraged Finance Conference, strategists pegged over $300 billion of AI-related high-yield and leveraged issuance globally this year, with US high-yield net volumes up 60% year-over-year purely on AI financing (At Any Rate, "Key Takeaways: European Leveraged Finance Conference 2026," September 17, 2026). On the investment-grade side, JPMorgan's credit team put data-center-related issuance near $300 billion with IG issuance up 50% year-over-year, calling data-center financing "the first inning" of the cycle (J.P. Morgan Making Sense, "Navigating credit markets in 2026," September 16, 2026). The eye-catching single deal: a 10-bank consortium providing a $22 billion loan to Blackstone and Alphabet's Crux AI cloud venture, backed by the chips themselves.

Nvidia's $14 billion Hugging Face grab raises flags. Nvidia agreed to buy the open-source AI platform Hugging Face for roughly $13–14 billion (against about $150 million of revenue, an ~86x multiple), framed as "chip demand insurance" against Google's and Amazon's in-house chips. Commentators expect antitrust scrutiny because it bolts dominant hardware onto a neutral software layer rivals depend on (Elon Musk Podcast, "Nvidia's 14 Billion Dollar Hugging Face Acquisition," September 3, 2026; Squawk on the Street, "Nvidia Buys Hugging Face," September 3, 2026).

Retail trading is booming, and Robinhood is the clearest window. CFO Shiv Verma described a business at roughly $5 billion of annualized revenue and ~$3 billion of EBITDA, with 13 separate business lines each generating over $100 million; the firm raised $2.2 billion in zero-coupon convertible bonds (no dilution until the stock more than doubles) and put $300 million straight into buybacks the same day. It's now #1 in retail options and #2 in equities (CFO Thought Leader, Robinhood CFO Shiv Verma, August 30, 2026).

Wall Street wants its own stablecoin. A consortium of major banks, including Goldman Sachs, Bank of America and Citi, is building a US-dollar stablecoin targeted for the first half of 2027, to compete with Tether and Circle (Daily Crypto News, "Wall Street Wants Its Own Stablecoin," September 2, 2026; CRYPTO 101, "The TRILLION Dollar Stablecoin Opportunity," September 21, 2026). In parallel, the SEC introduced an "innovation exemption" for tokenized-stock venues just days after crypto's Clarity Act stalled in Congress, though adoption depends on whether real issuers embrace it, and some (like AMC's CEO) have pushed back (Unchained, "Crypto's Clarity Act Collapses. Two Days Later, the SEC Introduces Its Innovation Exemption," September 18, 2026).

Secondaries have become the market's "release valve." With IPOs and buyout exits still clogged, the private-market secondary business (where investors sell their fund stakes to others for cash) is now running at roughly $250 billion a year, up from $5–6 billion in the early 2000s. VanEck's data showed only 6 of about 1,000 "unicorns" reported down rounds in the past year, but roughly half haven't priced a new round in 2–3 years, leaving "low trillions" of stale private valuations waiting for a mark (Animal Spirits Podcast, "Talk Your Book: Why Aren't There More IPOs?," September 14, 2026).

A "compute futures" market is being born. CME Group is launching futures contracts on AI compute power, settled against an index of H100 GPU pricing, a spot market described as roughly $300 billion last year and projected toward $2 trillion this year (Energy Gang, "A new market for AI compute: What GPU futures could mean for energy," September 8, 2026).

The Fed hiked, and the reaction was a shrug. The Fed raised rates a quarter point (its first hike since 2023), but several podcast voices argued the move was largely symbolic against the bigger force of AI-driven growth and issuance. As one strategist put it, mortgage rates near 7% matter more to actual capital-markets activity than the Fed funds rate itself, with the 10-year at ~4.94% and 30-year at ~5.3% (The Dividend Cafe, "The Most Performative Interest Rate Hike Ever," September 18, 2026).