Most institutional venture models still assume a software company at $1M ARR needs $4–6M of cumulative capital to reach product-market fit. That assumption was correct in 2020. It is materially wrong in 2026. AI-assisted development has compressed the cost of building, iterating, and shipping production software by a factor of three to five for a meaningful share of categories. Allocators who have not adjusted their models are over-pricing risk and under-pricing capital efficiency.

The shift is structural, not cyclical. Code-generation tools, model-driven testing, and AI-orchestrated devops have moved from novelty to default in serious engineering organisations. A two-engineer team in 2026 ships at the velocity a six-engineer team did in 2022. Time-to-MVP has dropped from 12 months to 3–4 in vertical-SaaS categories. Time-to-paying-customer has compressed similarly. The result is not "cheaper companies" — it is companies that reach the same milestones with materially less capital, leaving more equity in founder hands at the moment institutional rounds close.

Three practical implications for allocators. First, ticket sizes need to come down. A $5M Series A in a category where $2M is genuinely sufficient creates a deployment problem disguised as a capital advantage. Second, milestone underwriting needs to shift from revenue-per-engineer toward revenue-per-dollar-deployed; the former rewards bloat, the latter rewards what the AI shift actually enables. Third, exit math needs revisiting — the same enterprise value can now be reached with smaller cap tables, which compresses the gap between best- and median-case outcomes.

The categories where this does not yet apply: hardware-heavy companies, regulated infrastructure, anything requiring deep clinical or compliance validation. There, the pre-AI cost structure still holds. But for the standard application-layer software business, the cost-of-progress curve has bent, and underwriting that has not bent with it is producing outcomes that look worse than they should.

At Brilwood, we are seeing this most clearly in the diligence packs we prepare for our clients — funds raising new vintages now lead with capital-efficiency multiples, not gross-deployment scale. Allocators have started repricing. Founders should make sure they are too.