One number reframes the current AI investment landscape: 65%.
According to Forbes, AI companies captured 65% of all venture deal value in 2025, up from 46% the year prior. That 19-point shift happened in a single calendar year. For context, AI’s climb from 46% to 65% of global deal value isn’t a rounding error or a statistical artifact. It reflects a fundamental reorientation of where professional capital goes.
The question this brief addresses isn’t whether AI is dominant in venture markets. It is. The question is what that dominance means for the companies that aren’t in the top tier, and whether the gap is structural or catchable.
What the Concentration Data Actually Shows
The 65% figure comes from a 2025 full-year dataset. It doesn’t map directly to 2026. But Morgan Stanley’s 2026 market analysis identifies AI as the central force influencing growth, earnings, geopolitics, and investment strategy across global markets, suggesting the trend isn’t reversing.
What the data does not show is distribution within the AI category itself. A sector that captures 65% of deal value can still be deeply unequal internally. Multiple market analyses suggest that the largest rounds are concentrating around a narrow set of companies, those with demonstrated computing access, global scalability, and the institutional relationships that attract infrastructure-scale capital.
The Nscale $2B Series C and AMI Labs’ $1.03B round, both covered in prior Tech Jacks Solutions reporting, illustrate the pattern. These aren’t Series A bets on unproven technology. They’re infrastructure-scale commitments to companies that can demonstrate they have, or can secure, the compute required to operate at frontier capability levels.
The Compute-Capital Fusion
The composition of top-tier rounds is changing in a way that the headline dollar figures don’t fully capture. Market analysts have observed that some recent mega-rounds bundle financial capital with computing resource commitments, equity combined with dedicated GPU access, cloud credits, or infrastructure partnerships.
This structure isn’t yet standard, and its prevalence isn’t quantified in available data. But the directional shift matters. If access to capital and access to compute are increasingly bundled together, then the selection criteria for a top-tier round have shifted from “who has the best technology” to “who can operate at the infrastructure scale required to compete.”
That’s a meaningful change for founders and investors alike. It implies that a technically superior product built on constrained compute may be structurally disadvantaged relative to a good-enough product with a committed infrastructure runway.
The Two-Tier Market
Industry observers describe an increasingly bifurcated market. On one side: a small number of companies with access to frontier compute, deep investor relationships, and the ability to attract capital at scale. On the other: the majority of AI companies competing for the remaining fraction of venture deal value under conditions that are more selective than the overall AI boom suggests.
The risk for founders and operators in the second tier isn’t irrelevance. AI applications built on existing infrastructure, APIs, fine-tuned models, workflow automation, continue to attract capital. The risk is mistaking overall AI investment growth for broadly distributed opportunity. The market is large. The top tier is narrow.
For investors, the bifurcation raises a different set of questions. Capital concentration in a small number of infrastructure-layer AI companies creates both efficiency (large bets on likely winners) and exposure (concentration risk if the infrastructure thesis doesn’t play out, or if regulatory action narrows the competitive landscape).
Strategic Implications
Three observations for market participants:
First, the 65% figure is a category-level statistic, not a company-level opportunity signal. Being in AI is not sufficient for accessing the top-tier capital environment. The relevant question is which tier a company realistically occupies, and whether the path to the top tier is open.
Second, compute access is emerging as a filtering criterion earlier in the fundraising process than it has been historically. Founders should expect infrastructure questions, what compute does the business require at scale, and how will it be sourced, to arrive in early diligence conversations, not just late-stage growth rounds.
Third, the geographic concentration of AI capital is worth tracking. Crunchbase data confirms large rounds in the UK and Paris, and the Morgan Stanley analysis references global investment dynamics, but the center of gravity for frontier infrastructure capital remains heavily concentrated in a small number of markets. Geography may become a compounding factor in access to the top tier.
The pattern across recent pipeline cycles, Nscale, AMI Labs, and now the broader trend data, is consistent. AI capital is concentrating. The companies it’s concentrating around share structural features that go beyond the quality of their models.
That’s the story the 65% figure doesn’t tell on its own.