BOT-05: Strategic Allocation in Skild AI

TLDR

A proposal to allocate 500,000 USDC from the DAO’s treasury to acquire equity exposure in Skild AI, a U.S. company developing a large-scale foundation model for robot intelligence. Skild AI is building the “mind” of general-purpose robots, an AI system capable of learning and executing open-ended physical tasks in the real world. This allocation provides the DAO exposure to the cognitive layer of Physical AI.

Skild AI’s founding team includes leading roboticists and AI researchers from Google DeepMind, Stanford, and Tesla Autopilot. Backers include Lightspeed Venture Partners, SoftBank Group, Jeff Bezos (Bezos Expeditions), Sequoia, Carnegie Mellon University, and Amazon’s Industrial Innovation and Alexa Funds. The company’s Series B-1 financing is being led by SoftBank, underscoring its strategic importance and positioning it among the most well-capitalized AI firms in robotics.

Exec Summary

Skild AI is developing a foundation model for robotics that integrates language, vision, and action to enable robots to perform complex physical tasks in real-world environments. Unlike companies focused on humanoid hardware, Skild AI is solving the cognition layer, the hardest and most valuable challenge in Physical AI.

Founded in 2023 by Deepak Pathak and Abhinav Gupta, Skild AI has already demonstrated robots performing multi-step household tasks from natural language instructions (cooking, cleaning, manipulation) as well as humanoids completing advanced locomotion (stair climbing, balance recovery, obstacle handling, load carrying).

The company’s Skild AI's Brain architecture is a large multimodal transformer trained on diverse datasets including video, sensor streams, and task demonstrations. This enables zero-shot generalization to novel physical tasks, a requirement for scalable deployment in homes and industrial environments.

With a world-class team and the backing of top-tier venture and strategic investors, Skild AI is emerging as a foundational intelligence layer in the Physical AI stack. This allocation gives the DAO exposure to a potential category-defining company before access narrows further.

Overview

Skild AI’s thesis is straightforward: robot hardware is converging faster than robot intelligence. As companies worldwide including Unitree, Agibot, EngineAI, Tesla, and Figure accelerate humanoid platforms, the bottleneck is cognition.

Skild AI addresses this with Skild Brain, a multimodal model designed to be hardware-agnostic. By providing perception, reasoning, and planning across robot embodiments, Skild Brain positions itself as the default operating system for robotics, similar to how Android became the standard for smartphones.

Rather than building physical robots, Skild AI focuses on cognition, control, and perception, the most defensible and scalable part of the robotics stack. Every Skild-powered deployment contributes data back to the model, enabling continuous improvement and shared learning across robot fleets.

During Skild AI's recent financing the company closed its Series B at a $4.2B valuation and is now raising an extension Series B-1 at an increased valuation that is not yet set.

Strategic Rationale

Humanoid robots are incredibly difficult to build. They must be versatile, safe, and able to operate reliably for long periods. Few companies are pursuing the full stack of both body and brain. Tesla and Figure AI are notable examples, while most others focus on hardware and rely on partners for intelligence.

Skild AI is positioned to be one of the standard cognition layers for humanoids, much like Android became the default OS for smartphones. Its Skild Brain, a foundation model designed to run on nearly any robot platform, has already demonstrated advanced capabilities across locomotion and manipulation.

By owning the intelligence stack, Skild AI could capture a large share of the value created by the humanoid boom even if it never builds a single robot. This aligns directly with XMAQUINA’s strategy to invest not only in humanoid robotics but across the full humanoid tech stack. If successful, Skild AI could emerge as the OpenAI of robotics, licensing its platform across industries from logistics and manufacturing to eldercare and domestic assistance.

The Deal

The DAO has secured access to participate in Skild AI’s Series B-1 financing through a SAFE (Simple Agreement for Future Equity) at a 10% discount to the B-1 valuation. The allocation will be made via a single-layer SPV (Special Purpose Vehicle), facilitated by a licensed broker. Upon closing of the B-1, the SAFE will convert into the same class of equity issued in that financing.

Fee Structure

1. Management Fee: A one-time 11.25% fee applied to the investment, paid directly to the independent SPV manager and broker. No portion of these fees are retained by the DAO, its members, or any internal unit. This fee covers legal setup, regulatory compliance, reporting, and ongoing administration.

2. Carried Interest: None. No performance fee is applied under any circumstances. The DAO retains 100% of the upside.

Transaction Details

-Allocation: 500,000 USDC -Instrument: SAFE with 10% discount into Series B-1 -Reference Valuation: Series B closed at $4.2B. Series B-1 valuation TBD at an increased level -Effective Entry Pricing: Approximately equivalent to Series B-1 institutional investors after SPV management fees. -Additional Transaction Costs: USDC to USD swap, off-ramping, and wire transfer fees expected to remain under $2,500 total

Potential Liquidity Events

Skild AI’s position as a foundation model company for robotics creates multiple paths to liquidity:

1. Initial Public Offering (IPO): If Skild AI continues scaling and establishes itself as the cognition layer across humanoid platforms, it could be positioned for a public listing within 3 to 5 years. Investor demand for AI infrastructure companies provides a supportive backdrop for such an offering.

2. Strategic Acquisition: Major technology and robotics players may see value in acquiring Skild AI to secure embodied AI capabilities. Potential acquirers include Amazon (already active in robotics through Kiva, Zoox, and Agility Robotics), Google/Alphabet (DeepMind and robotics initiatives), Microsoft (expanding in AI through its OpenAI partnership), NVIDIA (extending beyond hardware into robotics software), or Tesla (to strengthen its Optimus program).

3. Secondary Liquidity: Limited secondary liquidity may become available after a 12-month holding period, depending on the company’s performance and market conditions.

While no liquidity outcome is guaranteed, Skild AI’s institutional backing, rapid valuation growth, and technical progress improve the likelihood of an exit scenario. Any exit or liquidation would require a DAO proposal and governance vote.

Key Risks & Considerations

1. Technical Execution Risk: Developing a general-purpose embodied AI is an unsolved challenge. Success depends on breakthroughs in model architecture, training data, and real-world deployment at scale. Failure to achieve robust generalization could limit adoption.

2. Time to Market: Commercial revenue is unlikely before 2028. This is a long-term, high-risk investment that may require patience and sustained capital before meaningful returns.

3. Competitive Landscape: Skild AI faces competition from well-funded efforts at OpenAI, Google DeepMind, Covariant, Tesla, and Figure AI, among others. The pace of progress by incumbents could reduce Skild AI’s differentiation.

4. Capital Intensity and Dilution: Large-scale model training and data collection require significant ongoing fundraising. If future rounds occur at unfavorable terms, existing investors could face dilution.

5. Adoption Uncertainty: While humanoid robotics has strong momentum, the timeline for widespread deployment in homes and industry is uncertain. Regulatory, cost, or technical barriers could delay adoption.

Mitigating Factors: Skild AI’s founders bring over a decade of embodied AI research from Carnegie Mellon, giving the team deep technical credibility. Backing from SoftBank, Jeff Bezos, Amazon, and tier-one VCs provides both capital and strategic partnerships. By focusing on cognition rather than hardware, Skild AI positions itself as an enabler across multiple humanoid platforms, with early demonstrations already showing feasibility in real-world tasks.

Governance Execution

If approved, the DAO’s Execution Engine will:

1. Finalize terms and compliance with the intermediary.
2. Coordinate legal and treasury operations for the transaction.
3. Document the full transaction flow with a transparent paper trail for auditability.
4. Once the Series B-1 round is closed and the SAFE converts, disclose the final terms of the investment, including the number of shares acquired, final price per share, and valuation.

All treasury activity and asset positions will be tracked and documented for DAO-wide visibility and auditability on the DAO portal: dao.xmaquina.io

– The Northstar Council