After years as a venture analyst, I still watch funding patterns closely. Where money flows tells you something about collective beliefs regarding where value will be created.

The AI funding landscape in 2026 looks different from 2024. The hype money has retreated. What remains is more targeted, more thesis-driven, and more focused on practical applications than moonshots.

Here’s what I’m seeing.

The Foundation Model Correction

The massive rounds for foundation model companies have slowed. After OpenAI, Anthropic, and a few others raised billions, investors realized how few companies could compete at that tier.

Training frontier models requires capital that only the largest funders can provide. Most VCs have moved on to finding value elsewhere in the stack.

What’s still getting funded: companies building specialized models for specific domains where the general-purpose models underperform. Medical imaging models, legal document analysis, industrial inspection - places where domain expertise matters more than scale.

What’s not getting funded: “we’re training a better general-purpose LLM” pitches without massive backing. The foundation model layer has consolidated.

Application Layer Activity

The most active funding is happening in the application layer - companies using AI to solve specific business problems.

Vertical AI SaaS is the dominant theme. AI-powered solutions for specific industries: construction, legal, healthcare, logistics, insurance. These companies combine domain expertise with AI capability to build products that generalists can’t easily replicate.

The winners here tend to have deep industry knowledge first, AI capability second. Pure AI teams trying to learn industries are struggling against industry experts who’ve added AI.

Workflow automation continues attracting investment. Companies that use AI to automate complex business processes - not just chatbots, but systems that can handle multi-step workflows with real-world actions.

The bar has risen. Investors want evidence of actual deployment and measurable ROI, not just demos.

AI-native tools for developers remain popular. Code generation, testing, documentation, debugging - tools that make AI more accessible to developers. This category benefits from clear users (developers) and measurable productivity gains.

Infrastructure Bets

Below the application layer, infrastructure plays are getting funded:

Orchestration and observability for AI systems. As organizations deploy more AI agents and workflows, they need tools to manage, monitor, and debug them. Companies building the DevOps equivalent for AI are attracting attention.

Data infrastructure specific to AI needs. Vector databases, feature stores, data labeling platforms, synthetic data generation. The plumbing that makes AI applications possible.

Edge AI infrastructure - enabling AI to run on devices rather than in the cloud. Chips, compilers, frameworks for on-device inference. The latency and privacy benefits of edge AI are driving investment.

What’s Getting Harder to Fund

Some categories have fallen out of favor:

General-purpose AI assistants without clear differentiation. The space is crowded, the incumbents are powerful, and investors are skeptical about differentiation.

AI for consumers without clear business models. Consumer AI apps that don’t have obvious revenue paths are struggling. The bar for consumer investment is high.

Research-heavy plays without near-term commercialization paths. Pure research organizations are having trouble raising from traditional VCs. The appetite for long-term bets has diminished.

International expansion of concepts that work in the US. Investors are more skeptical of “we’ll take this US model to Australia/Europe/Asia” pitches than they were a few years ago.

Geographic Patterns

US funding dominates AI overall, but patterns vary by region:

Australia is seeing modest AI startup activity, with strength in specific verticals (mining tech, agriculture, healthcare) where local domain expertise matters. The ecosystem is smaller but the opportunities in AI implementation for existing industries are significant. AI consultants Sydney and similar firms are seeing increased engagement from local businesses seeking to adopt AI.

Europe has strong AI research and some funded startups, but struggles to create the very large companies. Regulatory considerations (AI Act) are shaping what gets built.

Asia varies by country. China has its own ecosystem largely disconnected from Western funding. Singapore and India have active AI startup scenes.

What Investors Are Looking For

Based on conversations and deal patterns, here’s what’s resonating with investors:

Evidence over vision. Investors want to see customers, revenue, or at least engaged pilots. Pure vision pitches are harder to fund than they were.

Defensibility. With AI capabilities increasingly commoditized, what’s the moat? Data advantages, domain expertise, network effects, regulatory approvals. Pure AI tech without moat faces skepticism.

Capital efficiency. The era of “raise huge amounts and figure it out later” has ended for most AI companies. Investors want paths to profitability that don’t require unlimited capital.

Talent with domain experience. Pure technologist founders are less compelling than teams that combine technical capability with industry knowledge.

Clear GTM. How are you going to sell this? Who’s buying? What’s the sales motion? Investor questions have gotten more practical.

Implications for Builders

If you’re building an AI company:

Solve real problems. The bar for “cool AI demo” funding has risen dramatically. Start with a business problem, not an AI capability.

Find your wedge. What’s the specific initial use case where you can win decisively? Broad platforms are hard to fund; specific solutions are easier.

Show evidence early. Design pilots and early deployments that generate compelling evidence. Investors want to see proof.

Know your customer. “Everyone” isn’t a customer segment. Who specifically buys this, and why?

Consider services hybrid. Pure product plays are harder to fund than products with service components that accelerate early revenue.

For organizations evaluating AI solutions, the funding patterns suggest which solutions are likely to persist. Companies with real customers and clear paths to profitability are better partners than those dependent on continued funding. Working with established AI consultants Melbourne may be more reliable than betting on heavily-funded but unproven startups.

The Cycle Continues

AI funding will continue to cycle. We’re in a more cautious phase after the exuberance of 2023-2024. This is probably healthy - the projects getting funded now are more likely to succeed.

The opportunities remain enormous. AI capabilities continue advancing. Business applications continue expanding. The correction isn’t about AI being less valuable; it’s about investors being more discerning about which specific companies will capture that value.

For founders and investors alike, the message is clear: practical applications with real customers and clear paths to value creation are what’s getting funded. The era of funding AI potential has given way to funding AI performance.