$222 billion continues to "fuel" the AI race, and capital is creating a batch of unprofitable zombie companies.

As large amounts of capital flow into the field of artificial intelligence, the market is facing an increasingly severe structural risk: huge sums of money are propping up a batch of startups that are fully functional but have lost their ability to survive commercially. These companies, unable to turn a profit or repay their debts, survive thanks to continuous financing "blood transfusions," and are referred to as "zombie companies" within the industry.

According to forecasts from the National Venture Capital Association of the United States, by 2025, venture capital deals in the AI and machine learning sectors will reach $222 billion, accounting for more than 65% of all US venture capital funding. This ratio marks a significant increase compared to 47% in 2024 and just 10% in 2015. However, despite abundant funding, product delivery speed cannot keep up with market hype, resulting in large amounts of capital flowing into companies that are in fact no longer competitive.

This capital misallocation has caused widespread concern among investors and economists. Landmark Wealth Management founder Joseph Favorito points out that long-term support for debt-financed or insolvent AI companies may slow down innovation and cause widespread economic ripple effects. This "blood transfusion" behavior crowds out limited resources that should have been allocated to more productive fields, leading to talent and capital being misallocated to companies with poor performance.

Market analysts warn that the current boom is similar to the period of the dot-com bubble. As the era of subsidies ends and industry consolidation intensifies, companies unable to generate real profits will face bankruptcy or being sold off cheaply, while those that can foresee this trend and possess genuine self-sustaining capabilities will emerge as the ultimate winners.

Capital Frenzy and Zombie Risks

In the AI startup sector, signs of "zombie companies" are becoming increasingly obvious. Economists define such companies as unable to repay debts, pay operating expenses, or generate sufficient revenue, but managed to survive by constantly injecting new capital, restructuring debt, or investors' unwillingness to accept losses. Normally, unproductive companies would collapse quickly once funding dries up, but amid the current AI boom cycle, abundant capital has extended their lifespan.

Data highlights the harsh reality of this market. Despite the frantic influx of funds, analysis by Demand Sage shows that around 90% of startups ultimately fail. Consulting firm Kearney reports that since 2010, the number of global "zombie companies" has grown by about 9% annually, and is expected to reach 2,370 by 2024. Fortune magazine’s recent report also noted there are already 574 such companies in the venture capital field.

Joseph Favorito states that nearly half of venture capital funds are flowing into the AI sector, enabling these companies that should have gone out of business to survive far past expectations. He believes the challenge for capital allocators is determining whether the funds are being wasted or whether they are likely to generate long-term returns.

The Hidden Real Economic Benefits

Industry insiders believe that cheap venture capital, government subsidies, and credit offered by cloud service providers mask the fragile fundamentals of AI startups. Abdur Rehman Arshad, CEO of Capidel Consulting, points out that these subsidies artificially keep corporate costs low for three to six years, thus concealing “true unit economics.”

Arshad predicts that costs related to AI will rise three- to ten-fold. He warns that by 2030, many companies will face a revenue gap of $800 billion, which will directly lead them to become "zombie" companies. Resume Tailor AI founder and former venture capital analyst Brian Londono holds similar views, saying many claims about AI "efficiency gains" are mixed with speculative growth and that funds are trapped in slow-growing companies. Once funds are released or subsidies stop, due to the prolonged mispricing of risk, these companies will face sudden devaluation.

Moreover, it is expected that by 2026, annual federal government funding for AI will reach $32 billion. Arshad believes that while this can extend companies' operating cycles, it may also prolong the survival of unsuccessful businesses.

Profitability Dilemma Amid High Costs

Besides capital misallocation, AI companies' high operating costs and unclear profitability paths are key factors behind "zombification." As pointed out by the Harvard Business Review, generative AI carries high variable costs and low variable returns, and the path to profitability for many companies is "uncertain."

Associate Professor of Business Administration Andy Wu analyzes that AI startups' initial costs are soaring, projected to increase from $50,000 to $2 million before seed funding by 2026. At the same time, productivity gains have not translated into revenue as expected. Research from MIT shows that 95% of AI application companies report no significant revenue growth. Arshad adds, early productivity declines are common, and this gap between inputs and outputs breeds many companies that rely more on subsidies than actual output.

Talent Competition and Market Shakeup

Loss of talent is another major survival threat AI startups face, making it difficult for them to translate technology into commercial value even if they possess it. Former economics associate professor Nikki Finley points out that if top talent is poached by large corporations, startups lose the chance to be acquired.

The Windsurf (formerly Codeium) case in July 2025 illustrates this. As the company was being acquired by AI startup Cognition, its leadership was poached by Google. Finley believes that while this may be a win-win for individual engineers and the overall economy, it also means that the independent value of the startup as an entity is hollowed out.

In the face of the current frenzy, Joseph Favorito compares this to the dot-com bubble period, when large sums of money were blindly invested in any project with ".com" in their name. He predicts that history will repeat itself, and as inevitable industry consolidation proceeds, those unable to adapt to the real market environment will ultimately disappear via bankruptcy or being sold off cheaply.

Risk Warning and DisclaimerThe market carries risks and investment should be cautious. This article does not constitute individual investment advice, nor does it take into account individual users' specific investment objectives, financial situations, or needs. Users should consider whether any opinions, viewpoints, or conclusions contained herein are appropriate for their particular situations. Investments made based on this article are at the user's own risk.