Early Stage Investing in the Age of Huge Pre-IPO Rounds

Key Concepts

• Early-Stage Venture Capital: Investing in founders and product vision at the "zero-to-one" phase before significant traction or revenue exists.
• Deep Tech & Robotics: The primary focus areas for Kindred Ventures' new $355 million fund.
• Inference Platforms: Infrastructure (e.g., Fowl, Base 10, Modal) that executes AI models, currently seeing massive growth due to token demand.
• Model Agnostic Harnesses: Software layers (like Perplexity) that integrate multiple AI models rather than relying on a single proprietary model.
• Embodied vs. Virtual Agents: The distinction between physical robots/autonomous vehicles and software-based AI agents for knowledge workers.
• Compute Constraints: The bottleneck where demand for GPU and data center resources currently outstrips supply.

---

1. Fund Overview and Performance

Kindred Ventures, a San Francisco-based firm, has successfully raised $355 million for a new fund dedicated to deep tech and robotics. This follows the success of their 2022 vintage fund, which grew from $200 million to a $1 billion gross fair market value as of June 2026, placing the firm in the top 1% of its class. Managing Partner Steve Jang emphasizes that despite the prevalence of massive late-stage, pre-IPO rounds, the "zero-to-one" early-stage investment remains critical for accelerating new technology frontiers.

2. The AI Ecosystem: Infrastructure and Application Layers

Jang identifies a clear shift in the AI market:

• Demand vs. Supply: There is a significant imbalance where demand for data center and GPU compute is far outstripping supply. This has fueled the growth of inference platforms like Fowl, Base 10, and Modal.
• The Role of Agents: The primary driver of current token demand is the rise of "agents"—both embodied (robots, autonomous vehicles) and virtual (knowledge worker assistants).
• Competitive Dynamics: A "battle" is emerging between:
  • Model Providers: Companies like Anthropic (Claude) providing state-of-the-art models.
  • Harness Providers: Companies like Perplexity that build multimodal, model-agnostic interfaces.
  • Agent Developers: Those building the most performant and productive applications for end-users.

3. Investment Strategy and Methodology

Kindred Ventures differentiates its approach from late-stage "mega funds" through a hands-on, founder-centric methodology:

• The "Zero-to-One" Framework: Unlike later-stage investors who focus on scaling existing revenue, Kindred focuses on product development, technology roadmaps, and team building before a product even exists.
• Symbiosis: Jang argues there is a natural symbiosis between early-stage firms (which provide the initial coaching and acceleration) and late-stage firms (which provide the capital to scale once the product-market fit is established).
• M&A vs. IPO: While the firm’s goal is for portfolio companies to remain independent and eventually go public, Jang acknowledges that the current market cycle will see significant M&A activity. Many legacy public companies are currently "catching up" to the AI cycle and will likely acquire smaller, innovative firms to accelerate their own transformations.

4. Key Perspectives and Quotes

• On the importance of early-stage investing: "Early stage is... a dwindling breed in this day and age of huge late stage pre-IPO rounds. But, we think it's absolutely not only still important, but actually even more important in this day and age of really fast growth."
• On the nature of the work: "We're investing in founders first, and we're investing in a product vision. They may not even have a product yet when we actually invest."
• On the market outlook: Jang anticipates a robust IPO market in the near future, citing companies like SpaceX, Anthropic, OpenAI, and Databricks as potential catalysts for market activity.

5. Synthesis and Conclusion

Kindred Ventures’ latest funding round underscores a strategic commitment to the foundational layers of AI and robotics. By focusing on the "zero-to-one" phase, the firm positions itself to capture value at the inception of new technologies. The core takeaway is that the AI market is currently defined by a massive demand for compute and a transition toward agent-based applications. While the firm aims for independent IPOs for its portfolio companies, it remains pragmatic about the role of M&A as legacy corporations scramble to integrate AI capabilities. The firm’s success is attributed to its specialized role as a coach and accelerator, distinct from the purely financial focus of later-stage venture capital.