AI Won't Kill Software, It Will Supercharge It

The software industry is grappling with a severe case of market anxiety. Since early 2026, public software ETFs have plummeted 30 percent, erasing gains made since the advent of ChatGPT. Giants like Salesforce, Adobe, and ServiceNow have seen their valuations drop between 25 to 30 percent in mere weeks, fueling fears of a 'SaaSpocalypse'. The prevailing narrative suggests AI will decimate the software sector.

This conclusion, however, misinterprets what software companies truly sell. The market is mistakenly treating software as a mere commodity, focusing on code rather than the deep value embedded in workflows and customer relationships. As detailed in a post from Andreessen Horowitz, the value has never resided solely in code; if it did, these companies would have been outcompeted by open-source alternatives or cheaper labor years ago.

Arguments predicting AI's destructive impact often fall into four camps: foundation models moving up the stack, enterprises adopting 'vibe code' replacements, product breadth expansion leading to intense competition, or a flood of low-cost single-person startups. The idea that AI agents will disregard brand loyalty for pure cost efficiency also fuels this pessimism.

The Enduring Moats of Software

The bedrock of durable competitive advantage, as outlined in frameworks like Seven Powers, remains relevant. While AI will alter the friction associated with switching vendors, making it easier for agents to migrate data, this shift forces companies to earn customer loyalty through superior products, not just lock-in. This will ultimately foster better innovation and a healthier ecosystem.

Network effects, crucial for platforms like Salesforce and Figma, will only grow stronger. As more users and AI agents interact within these ecosystems, their value compounds. AI-native platforms are already leveraging this, connecting service providers, clients, and agents in self-reinforcing loops.

Scale offers advantages, particularly in AI applications where compute costs are high. Centralized infrastructure, like Stripe's compliance and payment optimization, benefits all clients by absorbing complex costs and improving with volume. Companies at the intersection of hardware and software, such as Anduril and Waymo, will also see unit cost reductions through increased production.

Brand loyalty, epitomized by the adage "no one ever got fired for buying IBM," will continue to matter, especially in business-critical functions. In a market flooded with new entrants, strong brands signal reliability and trust, a crucial differentiator that upstarts cannot replicate overnight.

Cornered resources, particularly high-quality proprietary data, will become even more valuable. While consolidating public data is becoming easier, AI unlocks new capabilities with unique datasets, making sources like Bloomberg's market data or Abridge's clinical conversations increasingly potent moats.

Process power, or 'process engineering,' represents perhaps the strongest moat. Application software encodes decades of institutional knowledge and workflow optimization. Companies like Harvey, which deeply understand specific firm processes, build an inseparable advantage that challengers cannot easily replicate, even with zero code costs.

Platform Shifts Create New Winners

Counterpositioning, where new entrants leverage business models unattractive to incumbents, will also drive disruption. The rise of AI agents is creating opportunities for startups to challenge established pricing models. For instance, companies like Decagon, pricing per conversation handled rather than per agent seat, offer a better incentive alignment that legacy players like Zendesk cannot easily adopt without cannibalizing existing revenue.

These platform shifts have historically created new market leaders whose models ultimately dwarf the old ones. AI is not fundamentally different; it represents a massive expansion of the software opportunity.

The future will see a bifurcation: AI will undoubtedly reshape vertical and functional software, but not through a massacre. Margins may find a new equilibrium, with AI boosting labor efficiency offsetting potential pricing power shifts. Crucially, the overall market size will expand dramatically.

The world remains software-short. As code becomes cheaper, demand will only increase. Companies will reach more customers, automate previously intractable workflows, and make low-ACV customers economically viable. Durable businesses with strong moats will not only survive but thrive in this dramatically larger software landscape.