Conference Presentation, Keynote
Scale or Stall: The ICONIQ 2026 State of AI | Seth Pierrepont & Vivian Guo | RAISE Summit 2026
- Iconic conducted a longitudinal survey of 305 AI executives (87% North American, 13% European) to benchmark the state of the industry as of May 2024.
- Legacy software companies now derive approximately 33% of total revenue from AI products, with projections reaching 50%+ by 2027.
- Average gross margins on AI products are expected to rise from 45% to nearly 60% by 2027, driven by model orchestration and cost optimization.
- Infrastructure AI products command an average gross margin of 67%, while consumer-facing AI products lag at 48%.
- Foundational model selection criteria have shifted: reliability and cost remain primary, but privacy, security, explainability, and enterprise SLAs have ascended in priority.
- Respondents utilize an average of 3.3 models per deployment, with over 50% relying on a mix of licensed APIs, open source, and proprietary models.
- Anthropic's market share as the top model provider surged from 51% to 81% over the last two quarters.
- Adoption of open-source models has grown significantly from negligible levels one year ago to widespread usage today.
- Pricing models are evolving away from pure subscriptions (currently 57%) toward consumption-based and outcome-based models, which now account for nearly half of companies.
- Outcome-based pricing is being adopted to align costs with value, particularly as AI agents allow one human to do the work of five.
- 50% of companies plan to scale their Forward Deployed Engineer (FDE) motions over the next 2–3 years, viewing the role primarily as a revenue driver rather than a technical stopgap.
- FDEs provide durable competitive advantage by transferring model capabilities into specific customer contexts and workflows.
- 66% of companies are prioritizing "agentic AI" capabilities for their product roadmap over the next 12 months.
- Development focus is shifting toward deep verticals (43% of builders) to address complex, regulated domains where specialized workflows are harder to dislodge.
- Financial services and healthcare/life sciences have risen to the third and fifth most targeted domains, respectively, as capital flows toward specialized AI applications.
- Revenue per Full-Time Employee (FTE) is projected to nearly double across both high-growth and non-high-growth companies within the next year.
- Largest enterprises (>$500M revenue) are adopting agentic workflows fastest, though productivity gains (20–27%) are uniform across company sizes.
- Human intervention remains critical in 95% of agentic workflows, primarily due to workflow breakdowns and lack of context for long-complex tasks rather than model hallucination.
- 68% of respondents cite difficulty in predicting the total cost of agentic workloads due to token spend, inference costs, and hidden integration/upskilling expenses.
- AI spend is projected to increase from 11% to 19% of total revenue, representing a shift from traditional 1% tooling budgets to significant human capital investment.
- Engineering coding assistance yields the highest productivity gains (32–48%), followed by legal review, data analytics, and marketing automation.
- High-growth companies ramp new AI tools 33% faster (2.5 months vs. 3.5 months) and utilize AI for a larger portion of their code generation.
- A significant adoption gap exists: while 71% of R&D employees have AI tool access, only 39% are "power users," with conversion rates lower in go-to-market functions.
- One-third of companies plan for a net reduction in headcount, though the majority are restructuring roles toward operational fluency rather than eliminating positions.
- AI-native organizations are flattening hierarchies, reducing layers between CEOs and junior individual contributors, and expanding spans of control.
- In R&D, non-engineers are increasingly shipping code features, with PR throughput increasing 25–40% and concept-to-prototype cycles compressed to under two days.
- Sales teams are utilizing voice agents for inbound qualification and closing AI-enabled coaching loops to improve ramp times and conversion rates.
- Finance teams are reducing close cycles from weeks to under one day, while HR and Ops are increasingly replacing human tasks with AI-first backfill.
- The prevailing strategy is no longer just acceleration of existing work, but the complete redesign of work structures to achieve operating leverage.