STOCK//SHORTERSUBSCRIBE — FREE

STOCK//SHORTER / THE THESIS

Software ate the world. Now AI is eating software.

The thesis in one paragraphMost software companies bill per seat, and a seat is a proxy for a human doing a workflow. AI agents now perform many of those workflows at a fraction of the cost — so customers renew with fewer seats, net revenue retention rolls under 100%, growth guidance gets cut, and the premium multiples that priced two decades of compounding compress violently. The equity reprices long before the revenue finishes declining. That repricing — not bankruptcy — is the trade.

1. The seat is the fault line

SaaS turned software into a tax on hiring: more support agents, more recruiters, more analysts meant more licenses. Wall Street capitalized that linkage as net revenue retention — and paid 10–15× forward revenue for businesses whose customers expanded on autopilot.

AI agents break the linkage. An agent that resolves most tier-1 support tickets doesn't trim a customer's seat count — it changes the slope of the headcount plan. Expansion cohorts become contraction cohorts with the same product, the same customer, and the same renewal date.

2. The multiple moves first

A vendor growing 25% at NRR 115% and a vendor shrinking 8% at NRR 85% can be the same company eighteen months apart. The first trades at a premium revenue multiple; the second trades like a melting asset. The stock-price difference between those states is routinely 60–80% — and it gets paid out before the income statement fully catches up, because markets price trajectories, not snapshots.

3. Where the exposure concentrates

  • Customer support platforms — agents already resolve the majority of tier-1 volume.
  • Legal research & discovery — paywalled-database moats versus models that read everything.
  • HR & recruiting software — screening and scheduling were the seats.
  • Outsourced IT & BPO — billing humans hourly for work models do in seconds.
  • Content & creative tooling — marginal cost of the output went to zero.

The ranked version of this list, with the signals we track per sector, lives on the AI disruption watchlist.

4. How the desk trades it

Structurally, not emotionally: defined-risk short structures on the most-exposed cohort, paired against the AI-infrastructure beneficiaries on the long side. The mechanics — instruments, sizing logic, hedges and their risks — are in the short-selling playbook. The daily evidence stream is The Daily Short.

Will software companies literally go to zero?

Most won't go to literal zero — but they don't need to for the short thesis to work. A SaaS stock priced at 12x forward revenue for compounding growth can lose 70%+ of its value simply by repricing to a low-growth multiple, while the underlying business is still operating.

Why doesn't better software protect SaaS vendors from AI?

Because the threat isn't a better rival product — it's the customer needing fewer humans. Seat-based revenue is a tax on customer headcount. AI agents shrink the headcount plan, so the vendor's revenue contracts even if its product never gets worse.

Which software companies are most exposed to AI disruption?

Highest exposure: seat-priced vendors whose users perform text-and-workflow tasks AI does well — customer support platforms, legal research, HR screening, outsourced IT/BPO, and content tooling. Lowest exposure: systems of record with regulatory lock-in, and infrastructure that AI itself runs on.

What's the bull counter-argument?

That incumbents embed AI themselves, charge for it, and turn agents into a new product line — converting seat revenue into usage revenue. It works for some. The short thesis is about the gap: most vendors will pivot slower than their customers cut seats, and the market prices the transition pain.

Read the unwind before it prints.

One free brief per week. The Daily Short blog every day. No spam, no positions without a plan.