Jun 19, 2025

Startup failure is not a rarity. In fact, it could be considered the norm, research suggests that between 50% and 60% of UK startups fail within three years, while other data suggests that fewer than 10% of startups can expect to celebrate ten years in business. While no founder sets out expecting failure, the harsh truth is that for many, their ambitions will go unrealised. Whether it’s because they couldn’t get the funding they needed, develop products quickly enough, or just fell victim to changes in the markets, startup failure is part of startup life. Transport startups that crashed Canoo, the electric vehicle manufacturer, was among the year’s first failures. The Californian startup planned to manufacture vehicles for fleets and ride-sharing, but its plans never came to fruition. Despite raising over $1 billion, having a multi-billion dollar valuation, and a partnership with established vehicle manufacturer Hyundai, a disappointing earnings projection in 2024 marked the beginning of the end. Despite announcing layoffs, the company filed for bankruptcy in January. While their Lilium Jet was innovative, a five-seat jet capable of vertical takeoff and landing, the company struggled to attain satisfactory performance and the necessary regulatory approvals for flight. After struggling for years, it declared insolvency after failing to raise the investment needed to continue. Pushing the limits of AI hype The AI trend is inescapable, and there are few products that aren’t seeing AI added, even where the benefits seem marginal. Builder.ai was not, it turned out, actually an AI company, but there is some suggestion that its financial failure was because there just wasn’t enough demand for AI-powered coding. They are not the only ones to discover that there is a limit to how much the AI hype can support. Rain AI promised to build the ‘platform that will power the future of AI’. Like other startups, it seemed well qualified to deliver on that. AI needs a huge amount of processing power, making hardware a limiting factor in the development of new models. The need for chips helped Rain AI win high-profile clients, with OpenAI signing a deal to purchase chips from them, and CEO Sam Altman investing personally while also helping them to raise investment. However, difficulties in prototyping and fabrication dampened investor enthusiasm, and changes in leadership failed to turn around performance. The company is limping on, but using bridging finance to maintain operations while it looks for a buyer, touting the value of their technology — not its chip-making potential — as its main asset. Healthtechs and agtechs that went hungry All startups seek to solve a problem, some sectors, like healthtech and agtech , tackle problems that can affect us all. But even that is no protection against failure. Spotlight Therapeutics was a healthtech at the cutting edge. A spinout from UC Berkeley and UCSF, it was developing ways to deliver CRISPR-based gene-editing therapies to patients. Had they been successful, the potential was huge. Funded by Google, they focused on ophthalmology, an area they determined had significant unmet need. The funding covered their research, but their initial results were underwhelming, resulting in the startup shutting down. Plenty Unlimited, however, showed that failure need not be final. The agtech company focused on indoor vertical farming, promising increased yields and pesticide-free products. With the world’s population is growing, and the area of farmable land is shrinking, it made the startup an attractive investment. Backers included Amazon’s Jeff Bezos and retail giant Walmart. By 2022, their last funding round, Plenty had raised $951 million. However, high energy costs and difficulties in scaling proved too much. But that is not the end of the story. Despite filing for bankruptcy this year, Plenty has managed to restructure and is now focusing on producing strawberries. While perhaps not the first produce people will think of when it comes to tackling world hunger, it illustrates that when it comes to startup failure, there can still be some green shoots of hope. A year like any other Although 2025 has seen some high-profile failures, it is by no means unusual. High failure rates are part and parcel of innovation. These failures provide lessons that other startups may want to learn. Whether it’s actually having the product you describe (Builder.ai), correctly assessing the regulatory hurdles (Lilium), or just making sure you can deliver on your vision (Rain AI and others), new startups can learn from the mistakes of others. One thing is certain: although we will see smarter and leaner startups emerging, there will never be a shortage of challenges to make life difficult for them.