Venture Capitalists Continue to Invest Billions in Generative AI Companies

The surge in funding for generative AI entities — those innovating AI-driven tools for crafting text, visuals, and audio content — continues unabated. However, this influx of capital is increasingly being directed towards a dwindling pool of early-stage companies.

During the first half of 2023, leading up to July 16, some 225 startups succeeded in securing $12.3 billion in venture capital, as per figures from Crunchbase shared with TechCrunch. Should this trajectory persist, generative AI firms are poised to either meet or surpass the approximate $21.8 billion amassed in 2023.

Funding dynamics for H1 2024 were elucidated as follows:

• 198 angel/seed transactions: $500 million
• 39 early-stage transactions: $8.7 billion
• 18 late-stage transactions: $3.1 billion

The major beneficiaries were early-stage initiatives, featuring Elon Musk’s xAI (garnering $6 billion in May), along with other notable startups such as China’s Moonshot AI ($1 billion in February), Mistral AI ($502.6 million in June), Glean ($203.2 million in February) and Cognition ($175 million in April). Chris Metinko, a seasoned analyst and reporter at Crunchbase, notes that investors are preferentially channeling funds towards ventures they believe have a significant chance of thriving, while lesser-assured early-stage projects are left to falter.

“A segment of VCs anticipates a deceleration in AI investment floodwaters, fueled by potential legal and regulatory hurdles AI entities might encounter domestically and abroad,” Metinko conveyed to TechCrunch. “Moreover, reflecting on the mobile boom over a decade ago, they surmise that the lion’s share of foundational infrastructure gains will accrue to already established tech behemoths.”

Echoing Metinko’s sentiment, the longevity of many generative AI startups, including those with robust funding, remains uncertain.

Generative AI frameworks are typically honed on publicly available text and imagery, with firms claiming protection under fair use in situations where this data is copyrighted. The eventual stance of the judiciary on this matter remains to be seen, pushing some companies to enter into licensing arrangements with copyright owners.

The quest for premium training data is getting progressively tougher and costlier as firms deplete available web resources and more content creators restrict data scraping. (An analysis forecasts the AI training data market ballooning from $2.5 billion to $30 billion within ten years.) Moreover, the training of models is not getting any less complex or affordable, evidenced by the $78 million and $191 million training costs of OpenAI’s GPT-4 and Google’s Gemini, respectively.

Given the hefty initial expenditures involved in developing leading-edge models, it’s hardly surprising that only a handful of generative AI startups are profitable — including industry giants like OpenAI and Anthropic. Reportedly, OpenAI, with an annual revenue nearing $3.4 billion, might face a $5 billion loss this year.

Nonetheless, investors in generative AI are evidently committed for the long haul, particularly those from leading tech entities like Google, Amazon, and Nvidia, who view these investments as crucial strategic initiatives. Yet, if generative AI startups fail to surmount their existential issues, the prospect of a bubble burst looms large.