We’re halfway through 2025. Time to look at what’s actually happening in startup land, beyond the headlines.

I’ve been tracking funding data, talking to VCs, and watching what’s getting traction. Here’s what stands out.

The Funding Picture

Venture funding in H1 2025 recovered somewhat from the 2023-2024 doldrums, but the character of the recovery matters more than the totals.

AI dominance continues. Roughly 40% of venture dollars went to companies with AI as a core component. This isn’t hype - it’s VCs betting that AI changes company-building economics enough to justify higher valuations even in a skeptical market.

Seed funding is healthy. Early-stage rounds are happening at reasonable pace. The “Series A crunch” that was a 2023-2024 problem has partially eased as some 2021-vintage companies either died or grew into their valuations.

Growth stage remains challenged. Late-stage rounds are harder to come by. Companies that would have raised $100M Series D rounds in 2021 are either not raising, doing flat rounds, or taking structure (liquidation preferences, ratchets) they wouldn’t have accepted before.

Bridge rounds are common. Lots of companies are doing small extension rounds to extend runway rather than full-sized rounds that require new valuation conversations. This kicks hard decisions down the road.

Where Capital Is Flowing

Beyond AI, some sector patterns are worth noting:

Climate tech resilience. Despite lower clean energy valuations in public markets, early-stage climate tech funding has held up. The thesis is that policy support (IRA in the US, similar programs elsewhere) provides durable tailwinds regardless of short-term market sentiment.

Defense tech momentum. Funding to defense and dual-use technology companies increased significantly. Geopolitical tensions plus perceived government spending increases make this sector attractive.

Healthcare AI. The intersection of AI and healthcare is drawing capital. Drug discovery, clinical decision support, administrative automation - these have clear value propositions and large markets.

Fintech consolidation. New fintech company formation has slowed. The money is going to established players expanding rather than new entrants. The low-hanging fruit has been picked.

Enterprise over consumer. B2B continues to dominate B2C in funding. The thesis: enterprise software has more predictable unit economics and more defensible positions than consumer apps.

Exit Environment

Here’s the less optimistic part.

IPO window barely cracked. A few companies went public, but the IPO market hasn’t truly reopened. Companies that hoped to IPO in 2025 are largely waiting.

M&A volume is up. Acquisitions are happening, often at prices that disappoint founders and early investors but reflect market realities. Strategic acquirers are bargain-hunting.

Private equity buying tech companies. PE firms are acquiring software companies, often taking them private. This provides an exit path but typically at lower multiples than growth equity would have supported.

Secondary market active. Employee stock and early investor positions are trading on secondary markets at meaningful discounts to last round valuations. This provides liquidity but crystallizes markdowns.

The practical implication: founders should plan for longer paths to liquidity than the 2019-2021 era suggested. Building for profitability and sustainability matters more when the exit timeline is uncertain.

Emerging Themes Worth Tracking

Some early-stage trends that could become important:

Vertical AI applications. Rather than horizontal AI platforms, I’m seeing more companies building AI deeply integrated into specific industry workflows. Legal AI, construction AI, logistics AI. These have clearer value propositions and go-to-market than general-purpose tools.

AI infrastructure and tooling. The picks-and-shovels companies serving AI developers: vector databases, model monitoring, evaluation frameworks, fine-tuning services. The thesis is that AI development will grow regardless of which specific applications win.

Cybersecurity for AI. New attack surfaces from AI adoption - prompt injection, training data attacks, model theft - create new security needs. Early companies are addressing these.

Human-AI collaboration tools. Not “AI replaces humans” but “AI makes humans more effective.” These seem to have better user acceptance and clearer ROI than full automation approaches.

Hard tech manufacturing. Companies building things in atoms rather than bits: advanced materials, novel manufacturing processes, hardware. Policy support for domestic manufacturing plus supply chain concerns make this attractive.

What VCs Are Saying (Privately)

The public narrative from VCs is always bullish - they’re selling. What they say privately is more nuanced.

Disciplined deployment. Most VCs I talk to are explicitly pacing their investments more slowly than 2021. They’re being more selective, doing more diligence, and letting valuations reset before deploying.

Tougher board conversations. Existing portfolio companies are getting harder feedback about metrics, burn rate, and path to profitability. The supportive hand-holding of the zero-interest-rate era is over.

Consolidation expectations. Many VCs expect meaningful consolidation in crowded categories. They’re thinking about which of their portfolio companies might be acquirers vs. targets.

Duration awareness. Funds raised in 2020-2021 have limited remaining investment periods. Some VCs are raising new funds; others are extending existing funds. This affects deployment pace.

AI skepticism beneath the enthusiasm. Despite funding AI heavily, many VCs acknowledge privately that they’re not sure which AI companies will actually build defensible positions. They’re betting on the space broadly while uncertain about specific winners.

Implications for Founders

If you’re building a company in this environment:

Unit economics matter from day one. The market for “growth at any cost” is gone. Build a business that could be profitable if needed.

Raise enough. In uncertain funding environments, raise more than you think you need. Running out of money with 6 months of runway is a bad negotiating position.

Focus on the wedge, not the vision. VCs want to see near-term traction on a specific problem before funding the grand vision. Get something working first.

Consider non-VC paths. Revenue-based financing, strategic investment, bootstrapping longer before raising - these paths may suit some businesses better than traditional VC.

Exit planning. Think about who might acquire you and what would make you attractive to them. This isn’t giving up on the big outcome - it’s risk management.

Implications for Innovation Managers

If you’re doing corporate innovation:

Startup partnership opportunities. Valuations have moderated, and many startups need customers more than they need VC dollars. Corporate partnerships, pilots, and commercial relationships are more attractive to startups than in the frothy years.

Acquisition opportunities. Good companies at reasonable valuations. If you have strategic acquisition appetite, this is a decent environment.

Talent availability. Layoffs at startups and big tech have created an unusual talent market. Building teams is easier than it was in 2021.

The startup ecosystem isn’t broken - it’s normalizing. The excesses of 2021 are behind us. What’s left is still dynamic, still producing interesting companies, just with more realistic expectations.

For serious participants - founders, investors, corporate partners - this environment rewards patience, discipline, and genuine value creation over hype and momentum.