Dan Ives’ $200M AI Fund Signals Investor Shift Toward Later-Stage Bets

The Wedbush analyst’s move into venture capital reflects a broader market correction: capital is consolidating around proven AI infrastructure plays, not speculative agent prototypes.

Dan Ives, the Wedbush Securities analyst who became a household name during the 2023-2025 AI hype cycle, has launched a $200 million fund targeting AI startups.

According to Benzinga, the fund will focus on “AI infrastructure and enterprise automation” — a deliberately broad mandate that contrasts with the niche bets dominating 2024-2025.

This isn’t isolated. Three other funds launched this week alone:

• SparkLabs and Mirae Asset’s Central Asia-focused fund for Series A+ AI startups (Pulse 2.0)
• MISUMI Americas’ hardware and physical AI fund (Supply & Demand Chain Executive)
• Project Ventures’ £5M Imperial College deeptech fund (Tech.eu)

The pattern is clear: investors are retreating from seed-stage agent experiments and doubling down on infrastructure layer startups with clear enterprise monetization paths. Ives’ fund is the canary — his public shift from analyst to VC capital allocator confirms institutional money sees the agent hype cycle peaking.

Why Infrastructure Over Agents Now

Ives’ fund prospectus reportedly emphasizes “scalable AI middleware” — the pipes, not the end-user applications. This aligns with two market realities:

1. Agent fatigue. The 2025 surge of prototype-level AI agents (AgentDock, Crew AI) flooded the market with undifferentiated chatbots repackaging GPT-5 wrappers. Enterprises now demand integration layers, not more standalone conversational UIs.
2. Hardware bottlenecks. As MISUMI’s fund demonstrates, physical AI (robotics, IoT sensor fusion) requires specialized components that pure-software startups can’t access. Capital is flowing to bridge this gap.

The SparkLabs/Mirae Asset fund’s focus on Central Asia is telling. Unlike oversaturated Western markets, regions with lower labor costs but strong engineering talent pools (Kazakhstan, Uzbekistan) are becoming hotspots for AI infrastructure development.

What This Means for Agent Developers

If you’re building yet another LangChain-based customer support bot, the funding window is closing. But three opportunities remain:

1. Vertical-specific automation. Project Ventures’ bet on Imperial College spinouts suggests deep domain expertise (biotech, materials science) paired with AI still attracts capital.
2. Hardware-software hybrids. MISUMI’s fund explicitly targets physical AI startups — a sector where agentic behaviors (autonomous drones, robotic pickers) require tight sensor-to-LLM feedback loops.
3. Legacy integration. Ives’ focus on enterprise automation implies that startups helping Fortune 500 companies retrofit existing workflows with AI will outperform those pitching greenfield agent replacements.

The days of raising $10M for a ChatGPT frontend are over. But as these funds prove, the AI market isn’t dying — it’s maturing. The next wave of viable agent startups will look less like BondAI and more like industrial middleware.

For builders, this means ruthless prioritization: either go deep on a niche with proprietary data/hardware advantages, or pivot to infrastructure. The agent directory shows which categories still have runway.