$100 million-plus funding rounds used to be incredibly rare. Now, 40% of seed and Series A rounds are clearing that bar

Early 2026 funding shows a concentration of very large early-stage AI rounds—Crunchbase finds >40% of seed/Series A investment has gone to rounds of $100M+—with recent examples including Humans& ($480M seed), Ricursive Intelligence ($300M Series A) and Merge Labs ($252M seed). Reports also cite mega-deal activity among incumbents (SoftBank in talks to invest another $30B in OpenAI after a prior $40B, and Tesla/Elon Musk contributing $2B to xAI’s ~$20B Series E), while Big Tech earnings were mixed this week (Meta revenue beat; Microsoft revenue growth slowed). The newsletter rounds up large venture financings (e.g., Decagon $250M; Oviva €200M/$239M; Flapping Airplanes $180M), PE buyouts and IPO filings (Liftoff Mobile targeting up to $762M; Ethos $200M offering), signaling both robust private-market liquidity for AI and elevated systemic risk concerns about valuation excess and potential correction.

Analysis

Market structure: Massive private capital into AI concentrates winners in compute, semiconductors (NVDA/TSMC), cloud providers (AWS/Microsoft/Google) and scaled ad platforms (META). Startups and later-stage loss-making firms face valuation risk as funding recycles into fewer, capital-intensive infra players; expect pricing power for GPUs and cloud compute to persist for 6–18 months, keeping gross margins elevated for NVDA and select cloud infra suppliers.

Risk assessment: Key tail risks are regulatory (EU/US AI safety/labeling rules within 6–18 months), a compute-supply shock (TSMC/NVDA capacity constraints over next 3–9 months), and a private-funding reversal causing steep markdowns in VC-held balance sheets within 12 months. Hidden dependency: most startups are hypersensitive to Nvidia GPU availability and cloud credit terms—an NVDA guidance miss is a system-level shock.

Trade implications: Favor concentrated, hedged exposure to NVDA (beneficiary of compute tightness) and select ad/engagement leaders (META) for 3–12 month horizons; short/underweight small-cap AI/uncashflowed private-to-public cohorts. Use call spreads on NVDA to limit spend and buy puts on small-cap tech baskets to hedge a funding unwind; buy/trim on 8–12% pullbacks, take profits at 25–30% rallies.

Contrarian angles: Consensus underestimates a compute bottleneck ceiling and overestimates survivorship—expect >40% of current large seed rounds to down-round or fail within 24 months. Historical parallel: late-stage internet funding surge pre-2000 shows high public correlation and sharp re-rating if revenue growth stalls; corporate cross-investments (TSLA→xAI) raise governance and capital-misalignment risk that can amplify drawdowns.

Key Decisions for Investors

• Establish a 1.5% portfolio long position in NVDA via a 3–6 month 10% OTM call spread (buy 10% OTM, sell 30% OTM) sized to 1.5% notional; hold 3–6 months, trim half if NVDA rallies >25% or if QoQ revenue growth <5% on the next print.
• Add a 2% long position in META (shares) with a 6–12 month horizon to capture ad-resilience; set hard sell if ad revenue misses consensus by >3% or if shares rally >30% from entry.
• Initiate a 1.5% short/hedge against small-cap AI exposure by buying 6–9 month puts on a small-cap tech/AI ETF or short a constructed basket of loss-making AI public names; cover if the basket outperforms the S&P by >15% over 3 months.
• Reduce net exposure to high-multiple cloud/software (e.g., MSFT) by 1–2% and, if retaining position, buy a 6-month 5–8% OTM protective put sized to cover holdings; unwind hedge if MSFT posts two consecutive quarters of revenue growth >5% vs prior-year.
• Require three leading indicators before adding size: NVDA 'days-to-ship'/backlog (must not shorten >15% QoQ), TSMC utilization (stay >92%), and absence of immediate restrictive language in draft EU/US AI rules; if any trigger breaches, cut discretionary AI exposure by 50% within 5 trading days.