2026-05-01-entree-capital-enterprise-ai-spend-map

Enterprise AI Spend Map (May 2026)

Source: raw/articles/2026-05-01-entree-capital-enterprise-ai-spend-map.pdf

Note: PDF supplied directly by user (not web-captured). Publication month is “May 2026” with no specific day — slug date set to 2026-05-01 by convention. Raw file renamed from the original Enterprise AI Spend Map.pdf on 2026-05-30 per user request; see refactor entry in wiki/log.md.

Summary

A first-party VC market map from entree-capital sizing 2025 enterprise AI spend at ~$84Bandforecasting$370–440B by 2030 (~5x, 35–40% CAGR). Decomposes the market into five categories (Foundation model APIs, AI-native apps, Embedded AI in SaaS, Self-hosted models, Fine-tuning) and traces dollar flows through to the labs, clouds, SaaS incumbents, and AI-native apps that capture the spend. Argues that by 2030 vertical AI + SLMs gain share while embedded AI in SaaS loses share — a thesis that directly tailwinds the team’s vertical-use-case-led-brain direction.

Key claims

• 2025 enterprise AI spend ~$84B,growing60–80$370–440B by 2030 (~5x at 35–40% CAGR). Anchored to a 34.3% blended GenAI software CAGR (~38% for enterprise alone). [PDF p.2, p.7]
• Three categories carry ~92% of 2025 spend: Foundation model APIs ($30B/36$25B / 30%), Embedded AI in existing SaaS ($22B/26$5B (6%) and fine-tuning $1.5B (2%) are the long tail. [PDF p.2, p.3]
• Foundation labs capture ~$35B(42$84B, directly via API revenue and indirectly via the token costs that AI-native apps (Harvey, Sierra, cursor, glean) pay back to the labs. [PDF p.2, p.4]
• **Where the $84Blands:\ast \ast Foundationlabs$34.75B (42%), SaaS incumbents margin $16.5B(20$16.25B (19%), AI-native apps margin $15B (18%). [PDF p.4]
• Labs monetize differently. openai is consumer-led: ~60% consumer, $15Bconsumersubsof$25B annualized (Feb 2026); 9M+ paid enterprise seats; API $1B+ARR(Jan26);Codex2M+weeklyusers+70$14B annualized of which $10BAPI+$2.5B Claude Code (>50% enterprise, doubled in 6 weeks Feb 26); reached $30B by April 2026. Google Gemini revenue not disclosed by Alphabet. [PDF p.5]
• Market is accelerating across all segments. AI-native apps +300% YoY ($6.25B\to$25B), Foundation APIs +140% ($12.5B\to$30B), Embedded AI +80% ($12.2B\to$22B), Self-hosted +70% ($2.9B\to$5B), Fine-tuning +110% ($0.7B\to$1.5B). [PDF p.6]
• Vertical AI is the breakout share-shift. Bessemer’s vertical-AI portfolio grew +400% YoY at ~65% gross margin. Scale VP and NEA frame vertical AI as targeting $11TUSlaborspendvs$450B US SaaS market. First vertical AI IPOs expected 2027–28 (Bessemer); legal, healthcare, and financial services named as the 2026–27 breakout categories. [PDF p.7, p.8, p.9]
• Embedded AI loses share. Gartner: 35% of single-purpose SaaS tools replaced by AI agents by 2030. Pie share drops from 26% (2025) to 20% (2030). Absolute dollars still grow because the total pie expands ~5x. [PDF p.7, p.8, p.9]
• SLMs become default; fine-tuning grows with them. Gartner: SLM use 3x larger than LLMs in enterprise by 2027. Self-hosted models share 6% → 10% by 2030; fine-tuning 2% → 4%. Domain customization becomes standard practice. [PDF p.7, p.8, p.9]
• Agentic AI = own category. Gartner: agentic AI overtakes chatbot spend in 2026–27; $752Bend-to-endby2029at119$70B ARR by 2028. [PDF p.7, p.8]
• Multi-model becomes default by 2026 to eliminate single-vendor risk (SaaStr/ETR survey). Anthropic enterprise share 21% → 48% (Sep 25 → Mar 26); OpenAI 62% → 56%; Google 27% → 40%. [PDF p.5, p.7]
• B2B grows faster than blended GenAI because consumer is closer to saturation. [PDF p.7]
• Methodology / scope. B2B vendor spend counted once at point of purchase; excludes hyperscaler AI capex (~$450B2026,paidbyhyperscalersthemselves),AIconsulting/services,andinternalAItalentpayroll.$84B = Menlo’s US enterprise figure ($37B) scaled to global by ~2.2x. Flow assumptions: ~70% of API revenue → labs / ~30% → cloud; AI-native SaaS pays ~30% of revenue back to labs as token costs; SaaS incumbents retain ~75% of AI add-on revenue. [PDF p.10]

Notable quotes

• “Foundation labs are the largest beneficiary — direct API spend + indirect via Harvey/Sierra/Cursor token costs.” (p.4)
• “By 2030: vertical AI + SLMs gain share; embedded AI loses share.” (p.2)
• “B2B grows faster than blended GenAI because consumer is closer to saturation.” (p.7)
• “Vertical AI explodes — Bessemer +400% YoY @ ~65% GM • targets $11Tlaborvs$450B software.” (p.8)
• “35% of single-purpose SaaS tools replaced by AI agents by 2030 (Gartner).” (p.8)

Related

• entree-capital — publisher
• vertical-use-case-led-brain — primary thesis this source validates externally
• vertical-ai-tam — the $11Tlabor/$450B SaaS framing as a standalone topic
• agent-native-go-to-market — agentic-AI-as-category data points
• clinical-data-portability — healthcare named as 2026–27 vertical breakout
• slm-default-and-fine-tuning — SLM 3x LLMs by 2027 framing
• foundation-labs-monetization-shift — the OpenAI vs Anthropic split this report makes explicit
• openai · anthropic · cursor · glean · sierra

Mentioned in