1. The New Geography of AI Power
2025 exposed a structural pattern no one can afford to ignore: AI is no longer geographically flat. Mega-round funding clustered overwhelmingly in three hubs, and each cluster demonstrated distinct liquidity profiles, exit dynamics, investor behaviors, and sector specialization.
The Bay Area continues to dominate with 32 mega-rounds (65% of all deals)—but the surprise breakout was Cambridge, MA, which captured six mega-rounds (12%) and emerged as a biotech-AI fusion ecosystem with a liquidity model fundamentally different from tech IPO pipelines.
Meanwhile, NYC (four deals) and Los Angeles (three deals) carved out specialized niches in finance/legal AI and media/generative AI, respectively.
Geography is no longer a backdrop.
It is a strategic determinant of how AI companies mature, raise, and exit.
As detailed in The State of AI VC (https://businessengineer.ai/p/the-state-of-ai-vc):
“Clustering is not an accident. It’s a signal that capital, talent, and sector-specific infrastructure have begun concentrating at an industrial scale.”
2025 is the clearest confirmation yet.
2. The Bay Area: Full-Stack AI Dominance
The Bay Area remains the gravitational center of AI with 32 mega-round companies, including:
- OpenAI
- Anthropic
- Groq
- Cerebras
- Scale AI
- Together AI
- Evenlabs
- Glean
This cluster isn’t just about density.
It’s about full-stack capability:
- labs (OpenAI, Anthropic)
- infrastructure (Groq, Together, Baseten)
- developer tools (Cursor, Cognition)
- applications (Glean, Replit-adjacent ecosystems)
The Bay Area is the only region with a complete AI supply chain—from frontier research to end-user applications.
The exit profile reflects this completeness:
- IPOs
- tech M&A
- strategic partnerships
- hyperscaler acquisitions
- horizontal platform expansion
It is the closest analogue to the “Silicon Valley of AI,” with capital recycling into the local ecosystem faster than anywhere else.
The key insight is this:
The Bay Area sets the direction. Cambridge defines the specialization. NYC and LA define the verticals.
3. Cambridge: The Biotech-AI Supercluster
2025’s breakout AI story is Cambridge, MA, which emerged as a biotech + AI hybrid ecosystem with unprecedented capital concentration.
Six mega-round companies anchor this new cluster:
- OpenEvidence ($400M+)
- Lila Sciences ($350M)
- Imelio Medicine
- BraceLife,
- Flagship-backed AI-biotech ventures
This is not a mini-Silicon Valley.
It is something different.
3.1 Why Cambridge Is Different
Three structural forces make Cambridge uniquely suited for biotech-AI convergence:
(1) MIT–Harvard Talent Density
Cambridge compounds three ingredients:
- deep ML expertise,
- computational biology leadership,
- translational research talent.
No other region combines these fields as naturally.
(2) Existing Biotech Infrastructure
Kendall Square is the world’s most concentrated biotech district, and AI snapped onto it like a missing piece:
- wet labs
- clinical trial networks
- pharma partnerships
- FDA regulatory expertise
- computational biology clusters
AI capabilities magnify existing industrial capacity rather than trying to build it from scratch.
(3) Flagship Pioneering’s “Company-Creation” Model
Flagship does not fund companies.
It designs them.
This aligns perfectly with AI-native biotech, where:
- proprietary data,
- proprietary models,
- proprietary biological targets,
- expensive wet-lab cycles
all require orchestrated company creation, not random founder experimentation.
Cambridge’s biotech-AI companies behave more like:
- hard-science R&D engines
- with AI augmentation
- financed by institutional capital
- and partnered with pharma from inception
This is the opposite of traditional tech startup logic.
4. Cambridge’s Liquidity Model Is Non-Tech
The most important insight for investors:
Cambridge AI-biotech companies exit on pharma timelines, not tech timelines.
This creates a fundamentally different liquidity profile:
Tech exits (Bay Area):
- IPO within 5–8 years
- fast M&A cycles
- revenue-driven multiples
- ARR-focused valuation
Biotech exits (Cambridge):
- partnership monetization
- milestone-based financing
- pre-revenue IPOs
- acquisition by pharma**
This bifurcates AI’s secondary market into two incompatible models:
- Tech-driven liquidity (fast, revenue-based, short cycles)
- Biotech-driven liquidity (slow, milestone-based, science-driven)
Investors need entirely different underwriting frameworks.
As summarized in The State of AI VC (https://businessengineer.ai/p/the-state-of-ai-vc):
“You cannot price Cambridge deals with Bay Area multiples. They follow fundamentally different physics.”
5. NYC and LA: Niche Clusters with Distinct Value
NYC: Finance + Legal AI
New York’s AI momentum is concentrated in:
- Harvey (legal AI)
- EvenUp (legal ops)
- finance AI + compliance + governance
- Wall Street partnerships
NYC’s value proposition is simple:
The largest concentration of rule-driven, document-heavy industries in the world.
The city optimizes for enterprise sales, regulatory tech, and compliance-heavy AI automation.
Los Angeles: Media + Generative AI
LA’s cluster is creative and generative:
- entertainment AI
- virtual production
- synthetic actors
- creative workflow automation
LA is where AI collides with:
- media,
- human creativity,
- IP law,
- Hollywood infrastructure.
Its exit paths resemble entertainment M&A more than tech or biotech.
6. The Strategic Logic of Geographic Clustering
The emergence of distinct clusters is not random.
It is a structural phenomenon driven by three forces:
(1) Talent Specialization
Regions develop deep domain specializations:
- Bay Area → AI research + infra
- Cambridge → biology + computation
- NYC → finance + regulation
- LA → media + creativity
This concentration amplifies itself.
(2) Capital Specialization
Investors cluster around the domains they understand best:
- Bay Area → AI + software
- Cambridge → biotech institutional investors
- NYC → fintech funds
- LA → entertainment investors
Capital markets reward local expertise.
(3) Liquidity Differentiation
Not all exits are created equal.
Different clusters generate different liquidity curves:
- Bay Area → fast liquidity
- Cambridge → milestone liquidity
- NYC → regulated-industry liquidity
- LA → content/IP liquidity
This has profound implications for secondary markets, fund strategy, and LP allocation.
7. The Meta-Insight: AI Is Becoming a Geographic Industry
As analyzed in The State of AI VC (https://businessengineer.ai/p/the-state-of-ai-vc):
“AI is entering its industrial phase, and industrial phases always cluster geographically.”
2025 marked the moment when:
- AI funding became spatially patterned,
- liquidity became domain-specific,
- and companies’ futures became shaped by where they are, not just what they build.
Geography is no longer incidental.
It is now strategy.
