VCs Should Focus on AI Applications: Silicon Valley Investor Zhang Lu

BEIJING, March 4 (AsianFin)—As the year 2024 kicked off, major U.S. tech companies successively unveiled their AI strategies, signaling the commencement of a new round of intense competition in the Artificial General Intelligence (AGI) field.

Based in Silicon Valley, Zhang Lu, as founding partner of venture capital Fusion Fund, maintains close connections with team members at OpenAI and NVIDIA. As such, have been witnessing the emergence of AGI from its nascent stage.

Since 2017, Fusion Fund has been investing in artificial intelligence (AI). By the end of 2021, Fusion Fund made a move into projects related to generative artificial intelligence. The investment in generative AI by Fusion Fund in 2023 exceeded the combined total of the previous years.

In Zhang's view, the early-stage investment ecosystem in Silicon Valley is very active, with high-quality projects quickly receiving oversubions, demonstrating a clear Matthew effect. Meanwhile, many startups no longer need to raise a lot of money because, with the application of AI, development costs are not as high, and efficiency has improved, reducing the need for hiring many engineers.

She believes that Sora is undoubtedly leading other text-to-video models on the market, and it has already demonstrated certain depth of understanding of the structure and rules of the physical world.

The phase of foundational AI innovation has passed, and the industry has now entered the application innovation stage. Zhang noted that 90% of consumer-facing AI innovation opportunities will belong to large tech companies, leaving only 10% as potential opportunities for startups.

For venture capital enterprises, opportunities are indeed not as abundant as before. Last year, the majority of funds invested in AI flowed into consumer-facing AI applications, but this inflow is expected to rapidly shrink this year.

However, it is worth noting that due to the disruption brought by AI, Zhang holds that the scale and potential of the entire market for industry upgrades and innovation are ten times greater than the internet era. Even if only 30%-40% of opportunities belong to startup companies, the overall market size and potential remain considerable.

In addition to understanding technological changes and VC logic, Zhang also observes that under the influence of generative AI driving industry changes, the organizational structure of modern companies will also undergo transformations.

In her view, the architecture of tech companies used to be pyramid-shaped, with relatively seasoned founders and CEOs at the top and many junior engineers responsible for product development in the lower layers.

However, due to the development of AI, company structures are likely to shift to an inverted pyramid shape. The upper-level founding teams will consist of more experienced individuals responsible for optimizing models and products, but there will be less need for many junior engineers in the lower levels.

The pace of model iteration is heavily reliant on the availability of large quantities of high-quality data. Companies with access to massive and high-quality datasets can optimize their models more effectively. Higher data quality may reduce the required training data volume, leading to lower computational costs, reduced power consumption, and increased efficiency, Zhang told TMTPOST.

This gives large tech companies a significant advantage in terms of computational power and quality of consumer data. Consequently, a major challenge for AI applications in the consumer sector is the dependence on who controls vast amounts of high-quality user data. However, such valuable data is mostly in the hands of major tech companies, posing a significant challenge for startups.

The emergence of platforms like Sora serves as a warning not only for video content startups but also for venture capitalists focusing on consumer-facing AI applications. Continuing investment in this direction may lead to continuous competition and challenges from large corporations, she noted.