In February 2026, $800 billion was erased from software stocks in five trading sessions. Jefferies called it the "SaaSpocalypse." JP Morgan titled their note "Software Collapse Broadens with Nowhere to Hide."
This wasn't a correction. This was structural repricing.
Anthropic released Claude Cowork — an agentic tool suite that automates legal, financial, and administrative workflows. The same workflows thousands of software companies sell as core products. Thomson Reuters plunged 16%. The iShares Software ETF hit levels not seen since the mid-2010s.
The market made a judgment: AI agents are compressing the value of the software application layer. And that compression is permanent.
But here's what the market hasn't figured out yet.
The Value Migration
When $800 billion leaves a sector, it goes somewhere.
Phase one was obvious: compute infrastructure. NVIDIA. The hyperscalers. Energy companies feeding data centers.
Phase two is less obvious: financial settlement infrastructure.
AI agents are becoming economic actors. They negotiate. They transact. They settle payments — sometimes millions of micro-transactions per day. Try doing that with a credit card.
Three constraints emerge when agents transact with other agents:
- Identity: Agents need blockchain-based proof of authorization. No human identity to inherit.
- Micropayments: Sub-cent transactions are uneconomic on card rails. Visa charges 1.5-3.5%. Agent settlement runs at 0.1%.
- Deterministic enforcement: Smart contracts encode settlement directly. No intermediaries. No disputes.
This isn't abstract. I run an AI agent. Axis — my chief of staff — helps manage investments, research markets, synthesize intelligence. Within months, I might fund a dedicated wallet for Axis with hard constraints on budget and scope. Let it proactively acquire datasets without waiting for my instruction.
That's Level 4 shading into Level 5 autonomy. And at Level 5, traditional rails break.
The Iran Catalyst
While the software collapse plays out structurally, a tactical catalyst is accelerating the rotation.
The Iran conflict has triggered a de-grossing event. Hedge funds were overweight "real world" assets — commodities, energy, shipping — versus "digital world" assets. Rate volatility spiked. Risk mandates unwound.
The result? The most hated trades of 2026 — software, Bitcoin, digital assets — are outperforming as shorts cover.
Andreas Steno at Real Vision put it bluntly: "We have turned upbeat on quality software, Bitcoin, and other high-beta tech proxies."
The conflict may resolve in weeks. Trump has a history of U-turning at peak escalation. But the de-grossing momentum continues regardless.
The Stablecoin Standard
Here's where it gets interesting.
Jamie Coutts, Real Vision's Chief Crypto Analyst, built a tracking system for the agent settlement thesis. Three variables, each sequentially dependent:
- V1: Stablecoins win as payment form factor — 73% confidence
- V2: Settlement happens on actual blockchains — 52% confidence
- V3: Public chains capture settlement — 43% confidence
Composite probability: 16.3%. Up from 4.7% fourteen months ago.
The evidence is broadening beyond crypto circles:
- Stripe integrated x402 for USDC payments
- Visa settled $4.6 billion in USDC on Solana
- PayPal launched its own stablecoin
- Mizuho's analyst note on Circle cited "agentic commerce" as a growth vector — first time a major bank referenced agent payments
Three of the world's largest payment companies are building on stablecoin rails. This is institutional adoption.
The Divergence
Not everyone agrees on positioning.
42 Macro's systematic framework says NO POSITION on Bitcoin and Ethereum. Their VAMS momentum signals haven't triggered.
Steno Research, running discretionary analysis, says INCREASED CONVICTION on Bitcoin and quality software.
Coutts is adding Circle (CRCL) — his first equity position — as a cross-scenario hedge on the stablecoin thesis.
The divergence reveals something important: systematic and discretionary lenses see different pictures. 42 Macro follows momentum. Steno and Coutts are betting on structural shifts before momentum confirms.
Both approaches have merit. The question is your time horizon.
What I'm Doing
I'm positioning for the structural shift.
The agent economy is real. I live it daily. My AI agent synthesizes the same reports I'm sharing with you now. It coordinates calendars, researches investments, drafts content. It will need payment rails that work without me in the loop.
That means:
- Bitcoin — Core position. Store of value. Liquidity.
- Ethereum — Settlement layer. Infrastructure.
- Solana — High-performance L1. x402 native.
- Circle (CRCL) — Cross-scenario hedge. Benefits whether public chains win or walled gardens dominate.
I'm avoiding:
- Individual SaaS names — Structural compression. The moat is gone.
- Asia equities — Hormuz vulnerability. Japan gets 90% of energy through the strait.
The Bigger Picture
We are at a convergence point.
AI agents are destroying intermediary value — SaaS, payment networks, insurance brokers, IT services. The $800 billion signal was just the beginning.
That same force is creating demand for settlement infrastructure. Infrastructure the traditional financial system can't provide.
Stablecoins are winning as form factor. Blockchains are emerging as rails. Public chains have a shot at capturing the autonomous agent economy — the slice that has no institutional counterparty, no compliance officer, no one demanding permissioned rails.
The market has priced the destruction. It hasn't priced the creation.
Position accordingly.
Marc Theiler
Kenai, Alaska