YC S26 Batch Data Shows Surge in Hard Tech Proportion
Vivian Midha Shen announced on social platform X a set of internal statistics from Y Combinator (YC) Summer 2026 batch (S26), showing a significant increase in the proportion of selected companies in the "hard tech" sector: the proportion of hard tech companies rose from 8% in previous batches to 20%, robotics companies from 1% to 6%, industrial manufacturing companies from 4% to 10%, defense-related companies from 1% to 5%, photonics and other computing hardware companies from 1% to 4%, and power infrastructure companies from about 1% to 3%.
This trend aligns closely with statistics from independent research organization Extruct AI based on 245 S26 companies: their statistics show that "industrial" companies, including robotics, manufacturing, hardware, energy, defense, and logistics, accounted for 23% of the S26 batch, setting a record high and nearly double the proportion of the previous two batches, significantly higher than the past five-year average of 6.6%; in contrast, the proportion of traditional SaaS companies has dropped to 12%, far below the five-year average of 28%, and the overall proportion of B2B software companies has fallen to 52%, the lowest since the S22 batch.
There is also a notable change in the founder structure: Vivian Midha Shen's data shows that in the S26 batch, 1 in 5 founders is a solo entrepreneur (compared to about 1 in 20 previously), and 1 in 6 founders holds a PhD. Independent statistics also confirm a significant increase in the proportion of solo entrepreneurs—19% of S26 solo founders, the highest level in the past five years (with a five-year average of 8.1%), of which 64% are engaged in software-related businesses, and 55% are under 25 years old.
Data also shows that the proportion of "end-to-end AI agent" companies in the S26 batch has reached one-quarter (previously about one-tenth), with typical scenarios including fully automated processes for medical billing and insurance claims. Meanwhile, independent statistics show that the proportion of companies labeled with "AI" in the S26 batch is 67%, down from the peak of 83% in the S25 batch, indicating that the AI label itself is transitioning from a "selling point" to a "default configuration," similar to how internet technology evolved from a selling point to infrastructure after 2005.
In terms of revenue performance, Vivian Midha Shen's data shows that the median monthly revenue achieved by S26 batch companies during the incubation period has risen to $20,000, significantly up from the previous batch's median of about $8,000, reflecting a clear acceleration in product validation and commercialization speed.
From the perspective of funding and competitive landscape, this round of "hard tech resurgence" essentially represents a shift in the focus of venture capital allocation: early-stage funding, which has concentrated on pure software and SaaS tracks for many years, is now rapidly shifting towards capital-intensive fields such as robotics, industrial manufacturing, defense, and power infrastructure, which were previously considered capital-intensive and had long return cycles. Beneficiaries are funds focused on early-stage hard tech investments and related upstream and downstream enterprises, while traditional pure software SaaS early projects are under significant pressure in competing for the same early-stage funding and incubator slots—evidenced by the drop in SaaS proportion to 12% in the S26 batch, directly reflecting this capital reallocation in the structure of incubator selections.
Independent statistics also show that the proportion of founders under 25 in the S26 batch has approached 60%, a significant increase from 38% in the previous batch, while the proportion of founders aged 25 to 29 has dropped to a five-year low of 23%; analysis suggests that these founders generally exhibit characteristics of "strong technical capabilities but relatively insufficient market awareness and market strategy experience."
Source: Public Information
ABAB AI Insight
As the world's most significant startup incubator, YC's historical shifts in batch structure correspond to turning points in the capital market cycle—after the 2008 financial crisis, YC increased support for enterprise-level SaaS, and around 2021, it embraced cryptocurrency and Web3 projects; now, the proportion of hard tech has jumped from single digits to over 20%, contrasting sharply with YC's earlier strategy of systematically rejecting pure hardware projects and focusing on light-asset software, marking a sort of "reversion" in YC's investment preferences after more than a decade.
The direction of capital movement is very clear: early-stage venture capital is shifting from the highly concentrated pure software and SaaS tracks of the past decade to capital-intensive fields such as robotics, industrial manufacturing, defense, and power infrastructure. The motivation behind this shift is that the maturity of AI technology provides a previously unavailable technical foundation for hardware automation and physical world automation, while geopolitical tensions (increased defense demand) and energy shortages (increased power infrastructure demand) have objectively opened new commercialization windows for these "heavy asset" entrepreneurial projects, allowing hard tech projects, once considered to have long return cycles and high risks, to regain early capital favor.
A comparable historical case is that after the mid-2000s, internet infrastructure gradually transitioned from a "core selling point" for startups to a "default configuration"—now AI technology is following the same path, with the proportion of AI-labeled companies in the S26 batch dropping from 83% in S25 to 67%, indicating that AI itself has transitioned from a differentiated narrative to a foundational capability. The current stage of the entrepreneurial ecosystem is a transformation phase where capital and technical capabilities begin to feed back into "physical world" (atoms) innovation, with robotics, defense technology, and energy infrastructure becoming the core narrative of a new entrepreneurial cycle.
This round of change essentially represents a restructuring of the industrial chain within the entrepreneurial ecosystem: for over a decade, the default configuration logic for early-stage entrepreneurial capital has been "light assets, high margins, and rapid scalability" software models; however, as AI technology has gained the ability to control physical systems and automate complex processes (such as the explosion of end-to-end agent companies handling medical billing and insurance claims), the previously clear boundary of capital efficiency between hardware and software has begun to blur—robotics and industrial manufacturing projects can also achieve rapid iteration and scalability similar to software projects through AI control layers. Mechanically, this explains why capital is willing to re-enter the hard tech field, once considered "too long a cycle and too low capital efficiency": because AI control systems are transforming physical world entrepreneurial projects into new species with software project-like growth curves.
ABAB News · Cognitive Laws
When a selling point is used for too long, it becomes a standard configuration.
The direction of capital reversion is always towards the fields that were neglected in the previous cycle.
The speed at which software transforms hardware determines the speed of capital return.