Since the incredibly influential development of generative artificial intelligence with OpenAI’s GPT engine, Silicon Valley has been swarmed by startups that promise to help the consumer using this new technology. Most of them are practically useless, with some of them just being money grabs for the founders. The excitement will die down soon, and with it many ventures will be swept away in companies that create unnecessary issues to solve.

AI startup funding reached 42.5 billion dollars in 2023, yet most of that money is wasted on impossible projects. The failure rate for startups over their lifetime is 90 percent, and AI startups building upon exorbitant investor interest have had a failure rate of 80 percent within just the last few years. Investors have a flawed perception of the abilities of AI, leading to misaligned goals with engineering departments. These misconceptions create unrealistic expectations for what companies can create and disappointment when the impossible expectations are not met.

Most AI products are queries to large language models (LLMs), programs that generate responses to questions based on analyzed data. These models are owned by larger companies such as Microsoft. LLMS cost millions of dollars to create, and most startups cannot afford to create one. Their dependence on other companies’ frameworks sets many up for failure, as their product has no real manipulation of AI tools and could be reproduced rapidly by any competitor. Investment into these companies creates shovelware, software that is primarily focused on quantity rather than its quality and therefore creates bloat. 

Around 42% of startups fail for the simple reason that their product is not practical. An example of this product is Humane’s AI pin. Intended to free users from the screens of mobile devices, the product was a total flop due to a combination of both underdeveloped artificial intelligence and giant goals that went unfulfilled. The investment in this company was based on baseless product promises and ultimately $240 million dollars of investors’ money went to waste. The willingness of investors to participate in gambles on unfounded technology doesn’t create the same potential jackpots as other startups, as the technology is continuously being rehashed and regulations are still being established.

Many startups chase the allure of investor capital, while failing to create useful products. AI is overpromoted and allows their founders to reap the rewards of the market while not creating real-world value. Olive, an AI company that specialized in healthcare, failed in 2023. The company pivoted 23 times into different medical fields and received more than 900 million dollars in investment—all while creating no actual usable product. This money could have been spent on more tangible research, but the allure of an “artificially intelligent” medical system was more alluring than that.

OpenAI has lost 5 billion dollars this year in expenditures, and almost all of its funding is derived from investment from Microsoft, giving it a safety net. Startups without the backing of a corporation are not expected to thrive in creating the infrastructure of artificial intelligence. 

Already, the technology boom of 2020 and 2021 is dying down. As we reach the limits of our current understanding of artificial intelligence, through LLMs, the idea of revolutionizing human lives with a promising new technology will become an unfulfillable hope for startups. 

Blossoming startups should let tech giants work with AI instead, as AI products are currently exorbitantly expensive to run and do not excel as consumer products. The startup bubble is about to burst, and investors who sink money into hype will be faced with the consequences of failed products, which wastes both human time and resources that could be better spent on other technologies.