Lightspeed Venture Partners has 603 portfolio companies tracked on the network, and all 603 publish APIs indexed here. Founded in 2000, multi-stage, roughly $25B+ across the platform, with Anthropic, Mistral AI, ElevenLabs, Wiz, Rubrik, Stripe, Affirm, Grafana and ngrok in the book.
Every company in the portfolio ships an API. Here is how they score.
The distribution
| Band | Count | Share |
|---|---|---|
| Exemplar | 13 | 2.2% |
| Strong | 27 | 4.5% |
| Developing | 85 | 14.1% |
| Thin | 72 | 11.9% |
| Emerging | 127 | 21.1% |
| Minimal | 278 | 46.1% |
| Unrated | 1 | 0.2% |
Forty companies out of 603 — 6.6% — score strong or better. Two hundred and seventy-eight score minimal.
The top of the book is genuinely excellent
| Company | APIs | Agent band | Score |
|---|---|---|---|
| Stripe | 159 | agent-native | 84.7 |
| Polygon | 12 | agent-native | 76.9 |
| Anthropic | 24 | agent-ready | 75.0 |
| Temporal | 7 | agent-ready | 73.0 |
| Grafana Labs | 138 | agent-ready | 72.7 |
Stripe at 84.7 with an agent-native band is close to the ceiling of what the rubric measures. Two agent-native companies and three agent-ready in the top five is a stronger agent posture than most funds’ entire books.
What the shape means
A power law in a venture portfolio is not a scandal — it is the model working as designed. A fund does not need 603 exemplars; it needs a handful of enormous outcomes, and Lightspeed has them.
But the distribution is worth reading as an operating fact rather than a judgement:
46% at minimal means the API is not the product yet. Minimal is not “bad API.” It is usually pre-product, stealth, or a company whose public surface is a marketing site with a contact form. For a seed-heavy multi-stage book, that is expected.
The 85 in the developing band are where the leverage is. These companies have shipped something real and have not yet invested in describing it. That is the cheapest scoring movement available in any portfolio — the gap between a working API and a well-described one is measured in weeks, not quarters.
The agent bands at the top set the internal benchmark. When a fund’s best-performing companies are agent-native, the portfolio has an in-house answer to “what does good look like” that no consultant needs to supply.
The comparison worth making
Bessemer — profiled later this week — tracks 519 companies with an almost identical top band: 13 exemplars. Its middle is slightly healthier (34 strong, against Lightspeed’s 27) and its tail slightly shorter (206 minimal against 278).
Two large multi-stage funds, two nearly identical exemplar counts, and the difference showing up entirely in the middle of the book.
Takeaway
603 portfolio companies, every one publishing an API, 6.6% scoring strong or better, and 46% at minimal. The top of the book is agent-native and world-class. The middle 85 are the ones where a quarter of description work would move the number.
See the full portfolio at apis.io/vcs/lightspeed-venture-partners/.