Deal Screening Card: Spatial Walk (北京空间步态科技)

Date: 2026-03-22 Analyst: AI Due Diligence Agent Stage: Series A (Raising) Sector: AI / Spatial Computing / Digital Humans

One-Liner

A 3D Gaussian Splatting (3DGS) platform focused on interactive digital humans, founded by top-tier CV academics (AD-NeRF/SadTalker), currently facing a "commodity trap" from Big Tech SDKs.

Scorecard

Dimension	Score	Signal	Justification
Team	7/10	🟡	Tier-1 research pedigree (S-Tier publications), but active professor status creates "FTE" and IP risks.
Market	4/10	🔴	Core 3DGS rendering is being commoditized by Meta/Qualcomm system-level SDKs.
Competition	3/10	🔴	Competing against "free" at the OS/Chip level; high pressure from 2D-gen leaders like HeyGen.
Product/Tech	6/10	🟡	Strong interaction/lipsync DNA, but original rendering moat has evaporated in the "SDK Era."
Financials	2/10	🔴	Extreme 100x P/S ratio ($120M Val / $1.2M Rev); critical 5-7 month runway remaining.
Risk	8.5/10	🔴	Inverted Risk Score. High technical dependency, funding cliff, and valuation-revenue disconnect.
Overall	4.2/10	🔴	High-Alpha academic bet with unsustainable financial metrics.

Key Strengths (Top 3)

1. Academic Pedigree: Founders are the creators of AD-NeRF and SadTalker, providing S-Tier credibility in neural rendering and animation.
2. Niche Technical Edge: Potential to lead in interactive 3D animation (lipsync/physics) where standard 3DGS viewers are static.
3. Early Hardware Optimization: Claims of running 3DGS on low-end ARM chips (RK3576) could offer a temporary window for B2B IoT applications.

Key Concerns (Top 3)

1. Valuation Bubble & Runway: 100x P/S is a "pre-2022" metric. With ~$700k monthly burn and <7 months of cash, the company is in a precarious "raise or die" position.
2. Platform Commoditization: Meta (Presence Platform) and Qualcomm (Snapdragon Spaces) are making 3DGS a free utility, removing the value of a standalone "engine."
3. Academic Entanglement: Potential IP conflicts with USTC/GBU and the risk that founders remain "part-time" researchers rather than full-time operators.

Verdict

Recommendation: Pass (Unless valuation is adjusted by >60% and IP is cleared). Reasoning: The project is "priced for perfection" while its core technology is being rapidly commoditized by platform giants. The burn rate vs. revenue trajectory suggests a high likelihood of a down-round or liquidity event within 2 quarters.

Questions for Founder Meeting

1. IP Ownership: Can you provide a formal IP transfer agreement or "Freedom to Operate" letter regarding the AD-NeRF/SadTalker delta?
2. Revenue Diversification: How do you transition from $1.2M (likely project-based) revenue to scalable SaaS/SDK revenue in the face of free Meta/Qualcomm tools?
3. Commitment: What is the transition plan for Prof. Guo and Prof. Cun to move to 100% FTE status at Spatial Walk?
4. Moat Definition: If I can use Snapdragon Spaces for free to render 3DGS, why would I pay Spatial Walk for a license?