Physical AI and Vertical Software Draw Heavy Capital in Late Q3 2026

Physical AI Hardware Attracts Institutional Weight

SiMa.ai closed a $150 million Series C at a $1.45 billion valuation, bringing its total capital raised to $500 million, as cxotoday.com reported. The round was oversubscribed and co-led by Fidelity Management & Research Company and Amplify, with AllianceBernstein, Baron Capital, and J.P. Morgan entering as new investors ^[1]. The breadth of that syndicate — spanning asset managers, growth equity, and a major bank — is worth noting. These are not typical early-stage venture names.

As YourStory reported, Dell Technologies Capital also participated, alongside StepStone Group and Alter Venture Partners ^[3]. SiMa.ai plans to use the capital to scale its agentic AI platform and develop new hardware products targeting humanoids, automotive systems, and drones. The combination of agentic software and purpose-built chips in a single platform is the architecture institutional backers are now willing to pay a unicorn premium to own.

PicoJool, a Palo Alto developer of vertical-cavity surface-emitting lasers, added a $27.5 million Series A to expand production of 200 Gb/s devices for AI data center interconnects, as optics.org reported ^[4]. It is a quieter deal than SiMa.ai's, but it points in the same direction: the infrastructure layer beneath AI — photonics, chips, edge inference — is drawing dedicated capital from investors who believe AI workloads will require hardware that does not yet exist at scale.

Proven Revenue Unlocks Series A for Supply Chain AI

Atomic, an AI-native supply chain planning platform founded by former Tesla supply chain leaders, raised $12.5 million in a Series A led by Klass Capital and Madrona, according to prnewswire.com ^[2]. The round is modest relative to the others this week, but the customer evidence attached to it is not. Atomic already automates 90 percent of purchasing for DoorDash's DashMart business and helped Good Chop cut its inventory by half while more than doubling revenue.

That pattern — a relatively small raise accompanied by striking operational proof points — reflects how seed and Series A investors are currently underwriting deals outside the largest AI platforms. Founders who can show a metric like inventory halved or purchasing fully automated at a named customer are compressing the risk investors need to price. The Atomic deal suggests that enterprise AI startups with tight vertical focus and documented customer results can still close institutional rounds without a unicorn-scale valuation.

Vertical Enterprise AI Reaches $4 Billion

EliseAI raised $350 million at a $4 billion valuation in a round led by Andreessen Horowitz and Bessemer Venture Partners, as financialcontent.com reported ^[5]. The company automates operations in housing and healthcare — two sectors with complex, high-volume administrative workflows that have historically resisted software displacement. Reaching a $4 billion mark puts EliseAI among the larger vertical AI businesses in the market.

The a16z and Bessemer co-lead is a signal in itself. Both firms have participated in multiple vertical AI rounds this year, and the EliseAI deal suggests their conviction in sector-specific AI platforms — as opposed to horizontal tools — has not softened. For a founder building in a regulated or operationally complex vertical, this is evidence that the investor appetite for that category remains active at meaningful scale.

What Connects These Deals

The rounds this week span pre-commercial hardware and revenue-generating enterprise software, but they share a structural feature: each company has a clear answer to where AI meets a physical or operationally constrained system. SiMa.ai targets inference at the edge in machines. PicoJool targets the optical interconnects those machines depend on. Atomic targets purchasing decisions inside supply chains. EliseAI targets the administrative layer of housing and healthcare. None of these pitches rests on AI as a general capability — each is anchored to a specific workflow or hardware constraint.

For founders raising a pre-seed, seed, or Series A right now, that specificity appears to be doing real work in investor conversations. The oversubscribed status of the SiMa.ai round ^[1] and the strength of the EliseAI syndicate ^[5] suggest that institutional capital is available but is concentrating around founders who can name the exact system their AI replaces or improves, and who can back that claim with either hardware in production or customer metrics.