The pace of investment in Gen AI firms, which are developing AI-powered products that generate text, audio, video, and other content, is not slowing down
However, they are being condensed into a decreasing number of early-stage ventures.
According to Crunchbase data shared with TechCrunch, 225 firms raised $12.3 billion from VCs from January to July 16 in the first half of 2023. If the trend persists, generative AI companies will equal or surpass the approximately $21.8 billion they raised in 2023.
The following is a breakdown of the H1 2024 total by stage:
$500 million in 198 angel/seed transactions
The total value of 39 early-stage transactions is $8.7 billion.
$3.1 billion in 18 late-stage transactions
The early-stage startups were the clear victors, including xAI (which raised $6 billion in May), Moonshot AI ($1 billion in February), Mistral AI ($502.6 billion in June), Glean ($203.2 million in February), and Cognition ($175 million in April). Chris Metinko, a senior reporter and analyst at Crunchbase, suggests that investors place bets on large ventures they believe have a high likelihood of success while allowing those they are less confident in to “wither away” during the initial stages.
Metinko informed TechCrunch that certain venture capitalists anticipate that the legal and regulatory obstacles that AI companies may encounter in both the United States and other countries will reduce the volume of AI funding. “Others argue that the most significant beneficiaries of the foundational infrastructure layer during the mobile revolution over a decade ago were well-established technology companies.”
To Metinko’s point, the future of numerous generative AI businesses, including those with the most substantial funding, appears uncertain.
Generative AI models are typically trained on data such as images and text derived from public web pages. Companies claim that fair use protects them from legal challenges if the data is discovered to be copyrighted. However, it is unknown whether the courts will ultimately rule in favor of generative AI companies, which is likely the reason why some have initiated licensing agreements with copyright holders.
High-quality training data is becoming increasingly difficult and expensive to acquire as startups exhaust the web’s supply. More creators choose to prevent crawlers from collecting their data, irrespective of the outcome of any single court case. (According to one analysis, the market for AI training data is expected to increase from $2.5 billion to $30 billion within the next decade.)
The process of training models is not becoming any simpler or more cost-effective, either: According to a recent Stanford report, the expense of training OpenAI’s GPT-4 was $78 million, while Google Gemini‘s cost was $191 million.
Despite the considerable upfront investment necessary to construct flagship models, few generative AI startups, including major players like OpenAI and Anthropic, are profitable. This is unsurprising. The Information reports that OpenAI, which is currently generating approximately $3.4 billion in revenue, may incur a loss of $5 billion this year.
Investors in generative AI appear to engage in a long-term strategy, notably significant technology investors such as Google, Amazon, and Nvidia, who regard generative AI investments as strategic bets. However, is it possible for the bubble to explode shortly? This appears to be a plausible outcome if generative AI firms cannot surmount the existential obstacles they are currently encountering.
