⏰ Only 12 days left until our Virtual DDVC Summit 23-25th March - Learn how Accel, Atomico, Bessemer, BlackRock and more use tools like OpenClaw, Claude, n8n or Harmonic to generate alpha
Brought to you by Harmonic - The Startup Discovery Engine
Founders Fund, IVP, Spark.
The world’s best investors use Harmonic to find their fund returners.
Welcome to another Data Driven VC “Insights” episode where we cover the most interesting research & reports about startups and VC from the past week.
Funding Rose 64% While Hiring Fell 10%: The Capital-Efficiency Inflection
Peter Walker, Head of Insights at Carta, analyzed 3 years of hiring and funding data across 50,000+ US startups on Carta, revealing a historic divergence where capital raised surged while headcount shrank - a pattern not seen since at least 2013.
+64% Funding Growth vs. -9.8% Hiring Decline (2023–2025): Total capital raised by US startups on Carta climbed from $78.44B in 2023 to $128.27B in 2025, while total hires fell from 329,202 to 296,854 over the same period; marking the first time funding and hiring have moved in opposite directions at this scale.
941,286 Hires and $312.5B Raised Over 3 Years: The dataset covers nearly 1 million hires and over $300 billion in capital across the full Carta universe, providing one of the most comprehensive views of the startup labor-capital relationship available.
Startups Truly Building with Smaller Teams: The divergence suggests a structural shift where startups are raising more capital but deploying it into AI tooling, infrastructure, and automation rather than headcount; fundamentally changing the relationship between funding and team size.
✈️ KEY TAKEAWAYS
This is arguably one of the most important charts in venture right now. For the first time in over a decade, startups are absorbing significantly more capital while hiring fewer people. For investors, this signals a new capital-efficiency paradigm where AI-native companies can scale revenue without proportional headcount growth, fundamentally improving unit economics and potentially compressing the timeline from funding to profitability.
84 Million Tokens in One Day: The Buy-vs-Rent Math Just Changed
Tomasz Tunguz, Founding Partner of Theory Ventures, shared his personal AI token consumption data, revealing that open-source models running locally on a $5,000 laptop now match December 2025 frontier intelligence; fundamentally altering the economics of AI inference for power users.
84M Tokens at $756 Cloud-Equivalent Cost in a Single Day: Running Kimi K2.5 via serverless API, Tunguz burned 84 million tokens on February 28th. At Claude or OpenAI blended rates of ~$9 per million tokens, that single day's work would cost $756 via cloud APIs, with peak days hitting 80M tokens and average days running 20M.
Subscribe to our premium content to read the rest.
Become a paying subscriber to get access to this post and other resources from our exclusive Data Driven VC community.
UpgradeA subscription gets you:
- Products like automation templates, prompt libraries, AI copilots
- 100+ masterclasses with experts from leading funds
- Access to our exclusive Slack community
- ... and lots more
