
The gap between depressed public software valuations and record-high private AI company valuations reflects a market entering a "show me" era, where startups must prove real ROI and embedding in enterprise workflows rather than simply claiming AI integration. Ranum of Sapphire Ventures sees 2026 shaping up as a historic IPO year, with major tech companies filing, and observes that actual enterprise demand for AI is concentrating in industrial settings like predictive maintenance, not in glamorous consumer-facing applications.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Sapphire Ventures partner Anders Ranum discussed how institutional investors are navigating divergent signals—public market software multiples at decade lows due to AI disruption concerns, while private AI startup valuations hit record highs. Software M&A activity picked up 40% year-over-year to $334 billion(約53兆円) across 678 transactions in 2025, though valuations have been reset.
Why it matters
Investors are entering a "show me" era where companies must prove concrete ROI and free cash flow rather than simply claiming AI integration. For enterprises, security, governance, and cost predictability have become deal-makers in AI deployments—making these line items defensible to CFOs skeptical of massive pilot bills. The bar for AI-native software has risen, but the opportunity remains real for companies that genuinely embed themselves in enterprise workflows.
What to watch
Ranum expects 2026 to be a historic IPO year, with SpaceX having gone public, Anthropic having filed, and OpenAI reportedly set to file soon. Below that tier, companies meeting higher profitability standards may wait until 2027 or beyond. Concrete enterprise demand is materializing in constrained industrial settings—packing, picking, inspection, and maintenance—where labor economics and clear buying cycles exist.
The article reveals a fundamental tension in the capital markets: public equity investors are pricing in disruption risk and marking down software valuations to decade lows, while private investors continue to bet at record valuations on AI startups. Ranum frames this gap not as irrational but as a real opportunity for investors who can distinguish between hype and genuine enterprise value. His 15 years at Sapphire and prior decade-plus at SAP position him to recognize the difference between superficial AI layering and fundamental workflow transformation.
The shift from a growth-at-all-costs ethos to what Ranum calls a "show me" era reflects a maturing market. Rather than stock bumps from AI claims alone, investors and buyers alike now demand evidence: free cash flow, clear paths to profitability, and proof that AI is solving real operational problems. For enterprise CFOs facing massive pilot bills, security, governance, and cost predictability have become non-negotiable—turning them from nice-to-haves into deal-closers. This raises the bar but also clarifies the playing field: companies that embed themselves into enterprise workflows and demonstrate ROI will win; those that merely layer AI onto existing processes will struggle.
Ranum's thesis on the fracturing LLM stack—where standalone companies like LangChain and WorkOS capture distinct, defensible layers—sits alongside his observation that consolidation is also happening. The distinction he draws is crucial: moats come not from being first in a category but from becoming genuinely hard to displace because the enterprise's actual processes run through the product. Meanwhile, the surge in industrial AI (his Tractian investment exemplifies this) reveals where near-term contracts are being signed: in labor-constrained, high-value settings with clear economics, not in the humanoid robots capturing Silicon Valley attention. This divergence suggests that practical, unsexy automation may deliver returns faster than glamorous moonshots.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack