Fireside Chat, Interview
a16z, Anish Acharya: Is SaaS Dead? Do Margins Still Matter? Why We Are Not in an AI Bubble?
- AI integration is expected to extend core business advantages rather than replace existing infrastructure like payroll or ERP, which constitute only 8% to 12% of enterprise spend, while incumbents like ServiceNow, Microsoft, and Adobe are predicted to improve their current workflows while startups win native categories such as AI-assisted movie creation.
- Market dynamics are shifting toward "outcome-based" pricing models, which may drag down traditional SaaS valuations, even as 75% of public SaaS companies have raised prices by an average of 8% to 12% (with some increases of 25% or more) since the release of ChatGPT, signaling strong product-market fit.
- The transition of software spend is anticipated to move from SaaS budgets to human labor budgets as coding and voice agents reduce migration costs, potentially growing the legal market from $50 billion to a range between 50% and 500% growth, including the emergence of AI-native law firms.
- Venture capital outlook predicts that companies will diverge rapidly, with competitors likely not overlapping in 12 to 18 months, while Andreessen Horowitz faces difficulty investing in directly competing firms due to conflicts over Fortune 500 buyers and engineers, though the firm aims to win 100% of deals it pursues in its specific domain.
- Investment criteria are evolving to prioritize "irrationally interested" founders and repeat entrepreneurs in enterprise sectors, while consumer markets favor beginners with a "beginner's mind," and Series A rounds are considered particularly difficult due to high valuations of up to 100 to 200 times revenue for companies with minimal product-market fit signs.
- Growth expectations vary by sector, with consumer markets allowing for "explosive growth" while ERP and payroll markets follow slower cycles, and the "tween round" of $500 million may be skipped if companies scale from $100 to $200 million in valuation with $1 to $3 million in revenue to $30 to $50 million within 12 to 24 months.
- Retention benchmarks have shifted, with Month 2 (M2) becoming the new standard for evaluating traffic after Month 1 (M1), where a 50% retention rate is considered solid and 60% to 70% is highly desired, driven by "inference as the new sales and marketing" for power users paying up to $300 per month.
- Technological evolution is predicted to create an aggregation layer for the application market where 80% of foundation models act as substitutes and 20% as specialists, requiring multi-model companies to provide rich feature surfaces that single-model providers like OpenAI and Anthropic cannot match.
- Future consumer interactions will likely favor "browse-based interfaces" over voice or dynamic UIs, with "contextual companions" serving as powerful tools for seniors and children to enhance pro-social behaviors, alongside "virtual dates" and "digital twins" that are viewed as under-hyped directions for the future.
- Investment risks are categorized into competitive, pricing, team, geographic, and fundraising types, with the speaker noting that while "inertia" is a powerful force making early leaders in 2023 and 2024 likely to maintain dominance, "agent maximalist" views of total automation are ahead of current capabilities, requiring human-in-the-loop exception handling.
- Productivity gains from automation are expected to manifest as a four-day work week rather than immediate job reduction, as high-ambiguity roles require human intuition, whereas BPOs and tasks with low ambiguity will see the most effective automation.
- Software spending is projected to asymptote to 80% to 90% of consumer discretionary spend by pushing the frontier of technology rather than cutting costs, with token costs on GPT-4.0 having decreased 100x since release, yet companies continuing to use closed models for quality and ambition maximization.
- Talent retention in London is viewed as potentially superior to San Francisco due to lower costs and reduced job-hopping, though San Francisco remains critical for accessing "shadowy hallway" secrets and requires a singular founder focus, while Tel Aviv's small domestic market forces immediate global expansion.
- Specific risks include the possibility of companies like Cursor losing 50% of revenue due to competitor cannibalization, the "SaaSpocalypse" questioning the durability of traditional enterprise revenue, and the "blended margin story" for AI-native companies being worse due to negative gross margin credits for trial users.
- The market is not considered to be in a bubble because supply capacity is consistently 100% spoken for with rising prices, unlike the overbuild seen in previous periods, and the speaker expects 100% of deals in their domain to be won, with price flexibility below certain thresholds but strict ownership terms to maintain partnership models.
- Founders must utilize products extensively to build intuition, as the investor landscape now demands direct product experience to evaluate opportunities like small business factoring, and the speaker anticipates entirely new categories emerging in 2026 as existing markets become saturated.
- Long-term predictions include the circumvention of execution constraints allowing individuals to fully pursue ambitions, the likely breakthrough in treatments for conditions like multiple cirrhosis, and the belief that automation will remove rote life tasks to improve the "NPS of the human experience."