Everything is the AI trade now

TL;DR

Four ostensibly separate April stories — SpaceX IPO, Blue Owl redemption crisis, Q1 bank earnings, Q1 M&A concentration — are four expressions of one thesis: that AI is going to reprice every knowledge-work revenue stream.
Public SaaS is down 22% YTD against S&P +4%. That's the anxiety ledger.
Private-credit NAVs are catching the same anxiety through ASC 820 marks. Insurance balance sheets hold 28% of private-credit AUM. Treasury is calling state regulators.
89% of Q1 US venture capital went to AI. 73% of mega-deal M&A value is AI-infrastructure-adjacent. JPMorgan's US$2.1B 2026 AI capex is disclosed as a specific line.
Five deals made the "M&A rebound." Strip them and the rebound falls to respectable-not-historic. One thesis carrying an entire cycle, in four asset-class costumes.
The base rate on single-thesis markets of this breadth is not comforting. That is not a timing call. It is a concentration warning.

The setup

Pick any financial story from April 2026 and ask what it's really about.

SpaceX files the largest IPO in history. On the surface, a launch-company listing. Underneath, a company valuing itself at US$2 trillion because it has consolidated xAI, because Starlink is the AI era's last-mile broadband layer, and because the US$1.4 trillion increment over launch-only valuations is an AI-infrastructure multiple.

Blue Owl hits its 5% redemption cap. On the surface, a liquidity event. In the fund's own investor letter, in writing, "heightened market concerns around AI-related disruption of portfolio-company revenue." The redemption is an AI anxiety expressing itself as a liquidity stress in the insurance-adjacent private-credit fund vehicle.

JPMorgan reports Q1 with a record trading quarter and an NII downgrade. The trading number was strong because of volatility produced, partly, by AI-exposed asset repricing. The NII downgrade is quietly about loan growth softness — in part because the borrowers are themselves repricing for AI risk. Dimon named borrower-revenue AI-cannibalisation as a top risk, explicitly, on the call.

Q1 M&A "Great Rebound." 73% of mega-deal value in five deals. All five are AI-infrastructure-adjacent. Strip them and the rebound is modest. Crunchbase: 89% of US venture dollars into AI.

Four stories. One thesis. Four costumes.

The thesis is: AI is going to reprice every knowledge-work revenue stream, and the market is already positioning for it.

The positioning is visible in public software multiples (repriced down). It is visible in private credit (repriced through borrower risk). It is visible in venture capital (concentration). It is visible in M&A (concentration). It is visible in bank capex disclosures (JPM's US$2.1B, named explicitly). It is visible in the largest IPO ever filed. It is, as of this month, the single most important factor in every major asset class.

Why this matters

The obvious response is: so what? Markets always have a dominant theme. That is true, and also not the right frame.

A dominant theme is normal. A single dominant theme that is simultaneously:

The reason public software multiples are compressed,
The reason private-credit funds are gating,
The reason insurance regulators are being called to Treasury meetings,
The reason mega-deal M&A is concentrating,
The reason venture capital is concentrating,
The reason bank trading revenue is high,
The reason bank NII is soft,
The reason the largest IPO in history is priced at a US$2 trillion valuation,
Inside the top risk factor on the earnings call of the country's largest bank,
And named by a private-credit fund in an investor letter,

… is not normal. It is the thesis having fully colonised the market's interpretive framework.

The historical analogue is not the dot-com bubble of 1999–2000. The dot-com bubble was concentrated in a single sector — technology equities. The AI trade is concentrated in a single thesis that is simultaneously expressing in multiple asset classes. In 1999, bonds, private credit, and bank earnings did not carry a tech exposure as their primary driver. In 2026, they do.

A better comparable is the 1997 Asian financial crisis setup — a single implicit bet (that Asian exchange-rate pegs would hold) priced across equity, currency, and credit. When the thesis broke, it broke across the entire asset allocation. Single-thesis cross-asset exposure is the structural risk of 2026. That is not a prediction that it will break. It is an observation that if it does, the conventional diversification playbook will not help.¹

The four threads, traced

Public SaaS as the anxiety ledger

Public software is down 22% year-to-date. S&P up 4%. The explicit investor concern, surfaced in Q1 earnings calls across the SaaS universe: will seat-based licensing survive AI-augmented knowledge work. The specific names bearing the weight are Salesforce, Workday, ServiceNow, Intuit, HubSpot. Not because they are most at risk but because they are largest and therefore most visible as repricings.

This is the ledger where AI anxiety gets recorded first, because public markets reprice fastest.

Private credit as the transmission

Private-credit funds lent heavily into mid-market software and business-services between 2021 and 2024. ASC 820 fair-value accounting requires those private loans to be marked against public comparables each quarter. Public SaaS multiples compress — private software marks compress with a one-quarter lag — private-credit NAVs compress — investors see losses — redemption requests hit caps. Blue Owl is the named case. The industry-wide number, US$20.8B in Q1 redemption requests across five major funds, is the less visible one.

This is the transmission mechanism — the pipe through which public-market anxiety flows into non-public balance sheets.

Insurance as the end-holder

State-regulated life insurers hold roughly 28% of private-credit AUM. This was a low-volatility trade through 2024. In April 2026, Treasury is asking state regulators to take a look. The worry is not default — charge-offs are normalising, not spiking. The worry is what happens to balance-sheet capital charges if the mark-down mechanism continues. NAIC's August 2026 meeting has a private-credit capital-charge methodology review on its agenda. The cover memo was drafted this month.

This is the end-holder — where the exposure compounds because it is illiquid, long-dated, and regulated on assumptions that were written in a different volatility regime.

Concentration in the up-trade

The flip side of the anxiety is concentration in the winners. 89% of Q1 US venture capital into AI. 73% of mega-deal M&A value in AI-adjacent deals. SpaceX at US$2 trillion. CoreWeave/Lambda at US$18B. The energy mega-deals (AES, Williams/Energy Transfer) priced explicitly on AI data-centre power demand.

The same thesis that is compressing one set of multiples is inflating another. Both movements are rational under the thesis. Both movements are the same movement. That is the signature of a monothesis market.

The Gartner-style split

Signal (what is genuinely new and likely durable):

AI has moved from an equity-only repricing story to a cross-asset-class thesis. This happened between Q4 2025 and Q1 2026 and is no longer deniable.
The transmission from public SaaS anxiety into private-credit NAVs is a real mechanism that was not present in previous tech repricings. It is the structural innovation of this cycle.
89% venture concentration in AI is a 15-year high in cross-sector venture allocation. The base rate on outcomes from such concentration is poor.
Insurance balance-sheet exposure to private credit is the slowest-reacting node in the system. It is also the node where the consequences are most regulatorily consequential.

Scepticism (what is overstated or inflated):

"This is 2000 again." No — the mechanism is different, the leverage is different, the regulatory frame is different. The shape is different even if the concentration is high.
"AI will cannibalise SaaS wholesale." Probably not. A more likely outcome: margin compression for seat-based licensing, absolute revenue growth for workflow and outcome pricing. The market is currently pricing the wholesale-cannibalisation case, which means the counter-case has asymmetry.
"Private credit is the next 2008." No — capital structure and gate mechanisms are working. The risk is asset-class repricing, not forced deleveraging.
"The SpaceX IPO is a top signal." Probably not — top signals are stocks with thin fundamentals. SpaceX has significant underlying revenue. The top signals, if they come, will be further out the risk curve.

What this means for you

If you are an institutional allocator. Your diversification playbook is partially compromised. An allocation mix of public equity / private equity / private credit / venture has a higher-than-advertised correlation exposure because all four have absorbed the AI thesis as a primary driver. Consider explicit thesis-hedged allocations (short software, long AI-infrastructure-light names) or increase allocation to assets genuinely orthogonal to the thesis (commodities, EM sovereigns, real estate with pre-AI lease structures).

If you are a corporate treasurer or CFO. Your bank lines, your public-market comparables, and your private-sponsor dialogue are all now conditioned by the AI thesis. Pricing conversations will assume a degree of revenue risk even if your business has no AI exposure. Pre-empt with data rather than argument.

If you are a board member. The strategy discussion in Q2 board meetings has to include an explicit AI-revenue-risk assessment, not because it's fashionable but because your lenders, investors, and M&A counterparties are already pricing it. A board that hasn't had this discussion by July is operating with outdated assumptions.

If you are an investor in a single asset class. Understand what you also own. A pure private-credit portfolio is an implicit software-cannibalisation bet. A pure growth-equity venture portfolio is an explicit one. A bank equity portfolio is a volatility-dividend bet with borrower-revenue-risk embedded.

If you are a policy-maker or regulator. The cross-asset monothesis is a systemic concentration that your existing frameworks — sectoral supervision, asset-class capital charges — do not measure. The measurement gap is at least 12 months wide. Worth closing before it becomes a hearing.

If you are a journalist covering markets. Every story this month is the same story. That is itself the story.

Uncertainty ledger

Revenue disruption realisation. Q2 and Q3 SaaS earnings will test whether AI-cannibalisation shows in actuals. If revenue holds, multiples recover and the thesis decompresses from its current pricing. If revenue softens materially, the thesis compounds.
Thesis breadth in Q2. If non-AI sectors start absorbing the AI thesis (through buy vs build, productivity uplift, margin re-expansion), the monothesis becomes a pervasive-thesis — which is differently risky. Watch consumer, industrials, and healthcare earnings for AI-adoption narratives that materially shift margins.
Central-bank policy. A Fed cut reduces refinancing risk for AI-exposed borrowers. A hold or hike accelerates the private-credit stress.
Regulatory action on private credit. Whether NAIC's August meeting produces a capital-charge adjustment is the fastest-moving regulatory variable. A 25bp increase in required capital changes the private-credit fundraising topology meaningfully.
The SpaceX IPO itself. June pricing is close enough to be a focusing event. A successful IPO (well-received pricing and aftermarket) confirms the thesis. A poorly-received IPO or a delay is a re-pricing moment.

Bottom Line

There is one trade this month and it is wearing four costumes. It is a public-software short, a private-credit gate, a bank trading tailwind, and a mega-deal M&A concentration — and they are the same trade. The diagnostic is not that this is about to break; it is that the standard diversification playbook no longer diversifies the way it claims to. Cross-asset correlation to a single thesis, at this breadth, has a poor historical base rate. That base rate is not a forecast. It is a warning that a portfolio labelled "diversified" in April 2026 is concentrated in ways the labels do not disclose. Read your own statements. The thesis is in all four of them.

Written in the tradition of — M.

Sources

Tier 1: SpaceX S-1 filing; Blue Owl Q1 2026 investor letter; JPMorgan Q1 2026 earnings call transcript; Federal Reserve H.8; NAIC 2026 agenda; EY-Parthenon Q1 2026 M&A Outlook; Crunchbase Q1 2026 Venture Report; SEC BDC filings (Apollo, Ares, Blackstone, Blue Owl, KKR)
Tier 2: Bloomberg; Financial Times; Wall Street Journal; Axios; AM Best private-credit AUM data; Pitchbook