Summary

A screen for beaten-down software names with positive free cash flow and revenue growth surfaced seventeen candidates from the "AI eats software" wipeout — but verifying every figure against SEC filings killed the two most eye-catching. Oracle is actually FCF-negative, pouring $56B into AI datacenters; AeroVironment's +133% "growth" is an acquisition, and it burns cash. What survives is seven genuinely profitable, cash-generative names down 40–79% on disruption fear, not broken fundamentals.

The premise, and the check that came after

A screen surfaced seventeen names from the software carnage: beaten-down prices, positive FCF, revenue growing. The premise is sound — when a narrative drives an entire sector down 40–79%, some real businesses get thrown out with the speculative trash. So we ran the screen, then did what screens can't do: verified every number against primary SEC filings.

That verification step killed two of the most interesting names on the list. Lead with that, because it's the honest story.

Oracle looked like the headline bargain. Down 58% from its high, with $31.98 billion in operating cash flow for FY2026. The screen loved it. Then we pulled the capital-expenditure line: $55.66 billion. Operating cash flow minus capex is free cash flow, and Oracle's FCF is roughly negative $23.7 billion for the year. That's not a cash machine — it's a company making one of the largest AI-infrastructure bets in the market, pouring capital into Oracle Cloud Infrastructure at a scale that dwarfs its operating earnings. You can own Oracle for that thesis. But you're not buying a quality-compounder at a discount; you're buying an AI-capex story, and the screen handed you the wrong reason.

AeroVironment looked even more interesting — down 58%, and the screen showed 133% revenue growth, defense tech exploding at triple-digit rates. Until you look at what drove it: the BlueHalo acquisition closed mid-period, inflating the year-over-year comparison. The combined company posted a GAAP operating loss of −$311 million and, more to the point, negative operating cash flow (−$78 million), FCF around −$141 million. The revenue growth is real but it was purchased, not earned, and the business is not generating cash.

This is why you verify. The screen alone would have put Oracle and AeroVironment at the top of your buy list — both for the wrong reasons.

What survives looks different

The names that cleared verification share one uncomfortable trait: they're genuinely profitable, cash-generative businesses the market has repriced on a thesis — that AI will disintermediate their core product — not because anything is operationally breaking.

A business printing 25–30% FCF margins while growing 14–30% is not a business being disrupted today. These are companies where the disruption is priced as a future terminal-value question: does AI eat the franchise in five years? If you believe it does, these prices still aren't cheap enough. If you don't — if Workday's HCM complexity, Atlassian's developer-workflow lock-in, and Zscaler's zero-trust architecture are durable — you're looking at genuine compounders down 40–67% on a narrative.

Atlassian deserves a specific note: it shows a GAAP operating loss (−$130M), the textbook "GAAP loss hides real cash generation" case. That loss is almost entirely stock-based compensation; the business itself generated $1.42 billion in free cash flow at 27% margins. The GAAP figure is the wrong lens here — the cash-flow statement is the right one.

The Trade Desk is the nuance — and the honest one

TTD is down the most, 79% from its high, with real FCF and real GAAP profitability. But growth has decelerated materially, from roughly 25% to roughly 12%, and Amazon's DSP buildout is a genuine competitive threat, not a narrative one. The market may be right to worry about TTD specifically. It's the one name in the cohort where the selloff has a fundamentals-based dimension alongside the fear — the line between bargain and trap. Own it only with conviction that programmatic advertising survives Amazon's encroachment.

Hims & Hers passes the screen on paper but barely — $57 million of FCF at a 2% margin while growth decelerates hard and capex rises to build fulfillment. That's not a quality cash business; it's a wellness operator in heavy investment mode. Worth watching as that capex cycle matures, not worth buying as an FCF story today.

The tape is the honest check on all of it

Nearly everything in this cohort is still in a downtrend. The fundamentals say quality; the chart says "not yet." Workday, Zscaler, UiPath, and Unity have shown early weak-up tape — basing, not confirmed reversals; Atlassian and HubSpot are still strong-down. Knowing a business is good is not the same as knowing the price has found a floor.

The right frame is a disciplined bargain hunt inside a disruption trade, not a bottom-call. If you're buying, you're taking a position on the AI-disruption thesis being wrong (or slower) for these specific names, sizing for the fact the tape hasn't turned, and entering with patience rather than urgency.

The screen gave us seventeen names. The verification gave seven worth considering and two cautionary tales. That ratio — and the process of getting there — is the actual product.

Sources

• All fundamentals (revenue, operating income, operating cash flow, capex, → FCF = OCF − capex) are from SEC EDGAR company facts (XBRL), latest fiscal-year filings for each issuer, pulled and cross-checked 2026-07-05. The screen's original figures were rejected where they disagreed with the filings (notably ORCL's FCF sign and AVAV's organic-vs-acquired growth).
• Drawdowns, trend, and RSI are from the desk's precomputed summaries (summaries), data as of 2026-07-02.