Is Software Dead? No. It Just Got a Lot Harder to Win. The SaaStr AI Deep Dive with Rory O’Driscoll

2026-07-15 · Source: SaaStrAI · Field: Business & Management — Entrepreneurship & Start-ups, Corporate Strategy & Leadership, Artificial Intelligence & Machine Learning · Depth: Intermediate, extended

Summary

Rory O'Driscoll, a 30-year software investor, asserts that software is not dead but winning has become significantly harder due to the massive AI investment cycle. In 2026, hyperscalers are projected to spend approximately \$688 billion on AI capex, yielding only about \$110 billion in revenue, indicating a half-trillion-dollar annual deficit. Revenue is not expected to surpass cumulative capex until 2031-2032, reaching around \$1 trillion, necessitating AI to capture 15-17% of the knowledge worker wage bill. The AI stack divides into "making AI" (chips, infrastructure, models) and "using AI" (applications). Defensible moats for "using AI" companies include combining software with sensors, leveraging marketplaces, utilizing proprietary non-public data, or adopting a full-stack business model. Pre-2022 software portfolios saw 10% become obsolete, while the remaining 90% split into insulated, additive, or threatened categories. Public software multiples have reset due to growth rate collapses from ~30% to ~10%, and compute intensity now varies wildly, from 70-80% of costs for model companies to ~10% for app companies.

Key takeaway

For founders and investors navigating the AI software landscape, recognize that the market demands clear differentiation and efficient resource allocation. Your product must either possess strong architectural or business model moats, or achieve hypergrowth by selling itself. Avoid token inefficiency and attempting to fund both heavy sales/marketing and high compute costs simultaneously. Critically assess your growth trajectory and category conviction, as pre-AI valuation multiples and growth expectations no longer apply.

Key insights

The AI industry is in a massive "invest mode," making software success harder but creating new opportunities.

Principles

• AI investment outpaces revenue by half a trillion dollars annually.
• Defensible moats are crucial for software companies "using AI."
• Software growth rates and valuation multiples have significantly reset.

In practice

• Identify your company's AI stack layer and moat type.
• Ensure growth rates match AI-native competitor benchmarks.
• Optimize LLM costs to ~10% of COGS for app differentiation.

Topics

• AI Investment
• Software Moats
• Foundation Models
• SaaS Valuation
• Compute Costs
• AI Applications

Best for: Investor, Entrepreneur, Director of AI/ML

Related on AIssential

• 20VC x SaaStr is Back!! Tokens Over Humans, the End of the SaaSpocalypse, and the Trillion-Dollar Land Grab — AI-driven automation and token spending are fundamentally reshaping business models, investment, and workforce structures.
• Is Software Dead? — AI capabilities are driving a market re-evaluation of the software industry, particularly SaaS business models.
• They spent MILLIONS on this... — Core functionalities of many SaaS applications are rapidly becoming replicable by AI agents at near-zero marginal cost.
• Ben Horowitz on AI Infrastructure, Economics and The New Laws of Software — AI fundamentally rewrites software's competitive rules, enabling rapid development but eroding traditional moats and demanding new infrastructure.
• AI Is Slowing Down — The AI industry's financial model is unsustainable, requiring trillions in revenue by 2030 that current demand cannot support.
• Rewriting Your Pitch: SaaS Isn’t Dead, But The Playbook For Founders Is Changing — The SaaS business model is evolving, demanding efficiency, defensibility, and outcome-based value in the AI era.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by SaaStrAI.