VC firms rarely reinvent themselves. Kleiner Perkins did—and has a new $3.5 billion to show for it

Kleiner Perkins has raised a new $3.5 billion fund in 2025, part of more than $6 billion raised across the Hamid–Fushman era. This follows 2024 vehicles including an $825 million KP21 early-stage fund and a $1.2 billion KP Select III, and the new capital is explicitly targeted at backing AI investments (e.g., Anthropic, Harvey, Applied Intuition). The raise signals a meaningful turnaround in Kleiner’s fundraising trajectory and increased venture capital deployment into AI-led opportunities.

Analysis

An influx of concentrated private capital into frontier AI lifts the entire upstream ecosystem rather than just the startups receiving checks. Expect demand for training cycles, bespoke accelerators, high-bandwidth fabrics, and observability tooling to grow 2–4x faster than end-market SaaS revenue over the next 12–24 months, squeezing supply chains for high-margin components and advantaging outsized-capex vendors. This creates a durable revenue tail for equipment and infrastructure names but also sets up cyclicality: when fundraising dries, utilization and multiples can swing sharply downward.

Competition dynamics will bifurcate: specialist infrastructure and IP-rich hardware suppliers win durable pricing power, while incumbents that try to buy growth through large follow-on rounds face margin erosion and talent arbitrage costs. The secondary effect is fewer IPOs and more late-stage private liquidity events, pushing return dispersion higher for LPs and making secondaries and tender markets more important signals of true pricing. Watch for contracting timelines on exits—realizable gains will be concentrated in 18–36 month windows, not immediate public-market re-ratings.

Key risks are macro tightening, a headline AI failure or regulatory shock, and a wave of down-rounds that force mark-to-market resets across private funds. Near-term sentiment can remain bullish for months, but a single high-profile model disaster or evidence of unit-economics failure in AI startups could compress valuations by 30–60% in 3–6 months. The most actionable contrarian: rotate from thematic software narratives toward capped-exposure infrastructure and selective long-duration optionality where cashflows are more predictable.

Key Decisions for Investors

• Long NVDA (6–12 month): buy January 2027 calls or 6–12 month delta-positive call spread — rationale: captures continued GPU/T4 demand; target +100–200% if datacenter cycle sustains; cap premium risk to <5% of portfolio value via spreads.
• Long AMAT + LRCX (6–18 month basket): equal-weighted core position in semiconductor equipment to play durable capex tailwinds — target 30–60% upside; set 20% trailing stop to protect against capex pullbacks linked to macro slowdown.
• Short a concentrated basket of high-multiple “AI-wash” SaaS names (example: PLTR and a second name sized as 1–2% each) vs. long small-cap AI infrastructure (SMCI) as pair trade over 12 months — aim for 2:1 reward:risk where a 40% decline in narrative names offsets a 25% gain in infrastructure; close if macro PMI stays >50 for >3 months.
• Allocate 3–5% portfolio to private-market secondaries or GP-led continuation vehicles with >1 year of verified revenue runway — objective: capture discounted liquidity and avoid future high-multiple entry points; limit exposure to avoid concentration risk and require third-party NAV verification.