AI Startups and the New Funding Cycle: What Founders Need to Know
AI startups are attracting an unprecedented share of venture capital. Here is what the new funding cycle means for founders, how investors evaluate AI companies, and how AI funding differs across regions.
Key points
- AI startups are attracting record venture funding.
- Investors focus on defensibility, data and unit economics.
- Competition and valuation risks are significant.
- AI investment is global, with regional strengths.
Artificial intelligence has become the dominant theme in venture capital. AI startups are attracting record funding, and investors are reshaping portfolios around AI. For founders, the opportunity is enormous, but so is the competition. Understanding how investors evaluate AI companies is essential.
The scale of AI investment
AI startups have captured a large share of venture funding globally. In Australia, Cut Through Venture and Folklore Ventures reported that startups citing AI attracted 61 per cent of total capital in 2025, around $3.1 billion.
Layers of the AI stack
- Infrastructure: compute, chips and data centres.
- Models: foundation and specialised models.
- Applications: products built on AI models.
- Tools: developer and data tools.
How investors evaluate AI startups
- Defensibility: proprietary data, distribution or workflows.
- Differentiation beyond wrapper products.
- Unit economics: compute costs and margins.
- Team expertise.
- Customer adoption and retention.
Risks for AI founders
- Competition from large technology companies.
- Rapid technological change.
- Compute costs.
- Regulatory uncertainty.
- Valuation bubbles.
What the funding cycle means for founders
- Opportunity: investors are eager to back AI companies.
- Competition: many startups pursue similar ideas.
- Higher bar: investors look for defensibility.
- Valuation risk: high valuations may be hard to sustain.
Regional perspectives
Australia. AI startups attracted most Australian venture capital in 2025, with activity across application, infrastructure and model layers.
United States. The US dominates AI investment, with large rounds for model developers and applications.
Europe. Europe has growing AI companies and regulatory leadership through the EU AI Act.
Asia. China, Japan, Korea, Singapore and India are investing heavily in AI.
A practical example
A hypothetical Australian legal technology startup built an AI tool trained on proprietary data from partner law firms. Investors were attracted by its data advantage, strong customer retention and focus on a specific workflow. It raised a Series A despite intense competition in AI applications.
Building a defensible AI business
As AI tools become more accessible, defensibility becomes the central question for investors. Common sources of defensibility include:
- Proprietary data that improves the product and is hard for competitors to obtain.
- Deep workflow integration that makes the product difficult to replace.
- Distribution advantages, such as partnerships or existing customer relationships.
- Domain expertise that allows the company to solve specific problems better than general tools.
- Network effects, where the product becomes more valuable as more people use it.
Founders who can explain their defensibility clearly are more likely to attract investors.
Managing compute costs
AI products can carry significant compute costs, which affect gross margins. Investors increasingly scrutinise unit economics, including the cost of serving each customer. Founders should track these costs closely and show how they will improve margins over time, through model optimisation, pricing or scale.
Regulation and trust
AI regulation is evolving. The EU AI Act introduces obligations based on risk levels, and other jurisdictions are developing their own approaches. Australia has consulted on guardrails for high-risk AI uses. Founders should consider regulatory requirements early, particularly in sensitive sectors such as healthcare, finance and employment. Demonstrating responsible AI practices can also build trust with customers.
AI for non-AI startups
Even startups that are not AI companies can use AI to build faster and operate more efficiently. Investors increasingly expect founders to use AI tools to improve productivity. This can reduce capital needs and extend runway.
Investor perspective
For investors, AI offers large opportunities but also risks of overvaluation and rapid obsolescence. Diversifying across AI layers and sectors, focusing on defensibility and paying attention to valuations can help manage these risks.
Questions investors should ask AI founders
Investors evaluating AI startups often ask: What happens to your product if the underlying model provider changes pricing or capabilities? How much of your value comes from your own data and workflows rather than third-party models? What are your gross margins today, and how will they change at scale? How do you manage accuracy, bias and safety? Founders who have clear answers to these questions stand out.
The importance of distribution
In crowded AI markets, distribution can matter as much as technology. Startups with strong channels to customers, through partnerships, existing relationships or a focused vertical strategy, often outperform those with superior technology but weak go-to-market. Founders should invest early in understanding and reaching their customers.
Staying adaptable
AI technology is changing rapidly. Startups that build flexible architectures, avoid over-reliance on a single model provider and stay close to customer needs are better placed to adapt as the landscape evolves.
Frequently asked questions
Do I need AI in my startup to raise money?
No, but AI is a major focus for investors. Non-AI startups can still raise with strong fundamentals.
What makes an AI startup defensible?
Proprietary data, distribution, workflow integration and strong customer relationships.
Are AI valuations sustainable?
Some may not be. Founders should be cautious about raising at valuations that are hard to justify.
What is an AI wrapper?
A product that adds a thin layer on top of existing AI models, often seen as less defensible.
Related reading
General information only. This article does not take into account your objectives, financial situation or needs, and it is not financial product, legal or tax advice or a recommendation. Figures are from the sources named and may change. Palo Alto Syndicate is a platform, not an adviser. It does not provide investment advice or recommendations. Every investment decision is yours. Early-stage investing is high risk. You could lose all the money you put in. See how the platform works and our terms of use.