CB Insights analysis of top 100 AI startups: the “most promising” 65 are US-based, though some have a dual HQ in China; overall US investment share is dropping
- Study finds companies focus on health, retail, transportation — Investment from other countries could slow American share
Context & Ripple Effects
CB Insights' ranking lands mid-arc in a running argument about where AI value accrues. Back in 2016, McKinsey counted US firms at 66% of global AI investment with China next at 17% (McKinsey's 2016 tally); by 2017, CB Insights itself reported Chinese companies had overtaken the US on startup funding, taking 48% of $15.2B versus 38% (the 2017 CB Insights funding report). The new top-100 list complicates both pictures: company formation still clusters in America, but the money behind it increasingly does not.
The dual-HQ detail is the tell — startups structuring themselves across both systems before geopolitics forces a choice. Later coverage sharpens the stakes: Preqin put US AI investment at $26.6B against China's $4B in mid-2023 (Preqin's 2023 comparison), yet by 2026 Bloomberg was tracking a narrowing gap driven by low-cost Chinese model releases.
First-order effects
- The 65 US-based companies on the list — concentrated in health, retail, and transportation — are drawing capital from a broader set of countries even as their incorporation stays American, diluting the US share of ownership in its own startup cohort.
- Dual-headquartered startups gain optionality now but inherit compliance exposure on two fronts, answering to regulators and investors in both Washington and Beijing.
Second-order effects
- Competing national capital pools respond by chasing the same verticals — health, retail, transportation — turning sector-specific AI startups into contested assets rather than purely domestic ones.
- Chinese AI startups responding to constrained domestic revenue by selling overseas, as MiniMax, ByteDance, and 01.ai did in launching products into the US market, meet these US-domiciled rivals on their home turf.
Third-order effects
- If dual HQ structures proliferate, corporate nationality becomes a negotiated attribute rather than a fixed one — forcing eventual regulatory sorting of which system a company actually belongs to.
- The pattern points toward AI leadership being measured less by where companies are founded than by whose capital, models, and markets they depend on — a dependency question rather than a headcount question.
The trend: AI startup formation remains US-centered while the capital and competitive pressure behind it globalize, making company geography an increasingly unreliable proxy for technological allegiance.