👋 Hi, I’m Andre and welcome to my newsletter Data Driven VC which is all about becoming a better investor with data and AI.
Upcoming events:
AI & Automation for Platform Teams Virtual Roundtable Sept 9th
DDVC Investor Summit at Bits & Pretzels 2026 Sept 28-30
Brought to you by Exa — The Search Engine Built for AI
Exa turns the web into live, structured data that your agents and pipelines can act on.
AI-forward firms use Exa to source companies by thesis, enrich target businesses programmatically, and monitor signals across the market in real time.
Sign up with code EXA50DATADRIVEN for $50 in free credits and bring real-time web data to your AI agents via Claude, Codex, Cursor & more!
Welcome to another Data Driven VC “Insights” episode where we cover the most interesting research and reports about startups, GPs, LPs, AI & automation.
Right-Sizing Your Intelligence Spend
Jaya Gupta at Foundation Capital argues that most enterprise AI workloads don't need frontier-level intelligence, and that the gap between lab incentives and enterprise outcomes is now measurable in the data.
320x Token Growth, Flat Returns: OpenAI reports that average reasoning-token consumption per enterprise organization rose roughly 320-fold over the past year, yet PwC's survey of 4,454 CEOs found 56% have not yet seen a significant financial benefit from AI.
7 to 10x Wasted Compute on Easy Tasks: Amazon researchers estimate reasoning models generate 7 to 10 times more tokens than necessary on simple tasks, giving the "overthinking" problem a concrete measurement rather than just an anecdote.
Efficiency Compressing Fast: Intelligence per joule improved 18x in just 16 months, and the 27B-parameter Qwen3.8 already matches or beats Opus 4.6 Max on several coding and agentic benchmarks.
✈️ KEY TAKEAWAYS
Labs profit from selling more intelligence; enterprises profit from needing less of it. Watch for internal routing layers and hybrid architectures to become the real infrastructure investment theme, ahead of frontier model access itself.
VC Ranking Flaws: Scale, Not Quality
Dan Gray at Odin stress-tests a popular VC ranking methodology by reweighting for capital efficiency, revealing how much "top 10" lists depend on how they're scored.
Top 10 Reshuffled by DPI-Weighting: Applying a weighted moving average of fund size to Ilya Strebulaev's ranking moved Union Square Ventures up 37 places and First Round up 36 places, among other double-digit swings.
New Leaders Emerge: The reweighted top 10 is led by SV Angel and Ribbit Capital, displacing several scale-driven names from the original "tier 1" list.
Methodology Is the Real Story: The exercise only reshuffles the original list of 50 firms; Gray notes a true top 10 by DPI would likely surface many lesser-known micro and seed funds instead.
✈️ KEY TAKEAWAYS
Most public VC rankings implicitly reward AUM and brand recognition rather than dollar-weighted returns. Before citing a "top 10" list, check whether it's scored on scale or on efficiency, since the two produce very different leaderboards.
Join 1934+ investors in our free Slack group as we automate our VC job end-to-end with AI. Live experiment. Full transparency.
Tender Offer Volume on the Rise
Hamza Shad at Carta shares a mid-year update on tender offer activity, showing that private-market liquidity mechanisms are scaling well beyond prior cycles.
71 Tender Offers, Up 34%: Carta administered 71 tender offers in H1 2026, up 34% from the 53 tenders administered in H1 2025.
$3B Transacted, Triple Last Year: Total transacted volume reached $3 billion, three times the $1.0 billion moved in H1 2025 and above the previous H1 record of $2.6 billion set in 2022.
Record-Adjacent Participation: Close to six in ten eligible sellers took part, and more than 90% of the shares buyers sought to purchase were successfully sold.
✈️ KEY TAKEAWAYS
Tender offers are becoming the default liquidity valve as companies stay private longer. The combination of higher deal count and near-record fill rates on both sides points to structural demand worth building into LP liquidity planning and employee retention strategy, rather than a temporary spike.
The Borderless Founder Advantage
Gabriel Vasquez, at a16z, lays out their playbook for backing international founders who keep one foot in their home market and one in Silicon Valley.
44% International Founders: 44% of portfolio companies in a16z's Apps early-stage investments feature international founders, split evenly between US and abroad HQs.
Home-Market Traction: A Stockholm seed company closed listed H&M and multi-billion-dollar Stena Metall as its first two customers; a Spanish startup converted a shared-university tie to Salesforce's CRO into its first major contract. Logo-to-stage ratio, not logo count, is the signal worth tracking.
Talent Arbitrage Is the Least Replicable of the Three Edges: One seed company hired six technical leaders who had each been CTO at a domestic unicorn, a pool priced far below Bay Area equivalents and invisible to US-only sourcing. Non-US unicorn alumni graphs and diaspora return-intent are indexable well before the market catches on.
✈️ KEY TAKEAWAYS
Diaspora networks are turning into a repeatable sourcing and customer-intro channel, not just a nice narrative. Funds should formalize relationships with home-market luminaries early, since those ties compound into deal flow and enterprise customers well before traction alone would justify them.
Upgrade your subscription to access our premium content & join the Data Driven VC community
Are VCs Actually Herding?
SVB’s State of the Markets H2 2026 report uses proprietary cap table and ownership data from Foresight to examine how often VC firms invest in the same companies.
Investor Overlap in the Top Five Nearly Tripled: 23% of known investors in the five largest US VC-backed startups have invested in two of the five, up from 9% in 2019. Similar concentration occurred in 2016 around Uber, Airbnb, Snap and WeWork.
Only 10% of Startups Show Significant Overlap: Across roughly 1,600 startups that raised funding since 2023, only 167 shared five or more institutional investors with another company in the sample.
74% Have Just One Top-10 Investor: Among companies backed by one of the ten most active investors, 74% have only one of those firms on their cap table. Just 6% have more than three.
✈️ KEY TAKEAWAYS
The most visible AI deals make VC look more concentrated than it is. When assessing managers, LPs should look beyond access to the biggest names and focus on where portfolios actually differ: company selection, ownership and entry timing.
VC Network Atrophy and Why Established Brands Move Later Stage
Ben Casnocha explains why a firm's brand and network can move in opposite directions over its lifecycle, and what that does to stage focus.
Networks Age as Brands Compound: As GPs age out or lose relevance, a firm's network can weaken even as its brand strengthens on the back of well-known early wins.
Brand Drives Deal Access; Discovery Still Needs Network. A strong brand helps a firm win competitive deals but doesn't help it find day-zero founders, which a fading network makes harder to source.
The Natural Drift Is Later Stage: Firms with strong brand and weaker networks gravitate toward later-stage investing, where brand recognition substitutes for early relationship access.
✈️ KEY TAKEAWAYS
Brand and network are distinct assets, and conflating them can mask a slow drift away from a firm's original stage focus. Emerging managers should treat active network-building as its own ongoing investment, separate from brand-building, or risk being pushed up-market whether the GPs intend it or not.
That’s it for today!
Stay driven,
Andre
PS: Join the AI & Automation for Platform Teams Virtual Roundtable 9th September
