Investors are excited about AI, pouring substantial funds into startups. U.S. venture capital funding soared to $55.6 billion in the second quarter, the highest quarterly total in two years. The surge is largely attributed to substantial investments in AI companies.

Over the long term, many industries are set to benefit from the technology, presenting growth opportunities that extend beyond giants like Nvidia and OpenAI. 

How are VCs strategizing to leverage AI, address its risks, and capitalize on emerging opportunities? We spoke with Andrey Belozerov, a leading international investor supporting AI companies across various verticals, including Big Data and cybersecurity.

The AI Venture Evolution

"We are still at the very beginning of the AI hype cycle," believes Belozerov. All the buzz and excitement have yet to be translated into useful tools and solutions.

AI projects often involve significant research and development. Seasoned investors like Belozerov advocate for long-term horizons, understanding that the most groundbreaking AI innovations will take time to mature and deliver returns.

"At the moment, tech giants, such as Microsoft and Google, are driving technological advancements," Belozerov said. "Yet, I envision a paradigm shift. There will likely be companies that will produce neither hardware nor software. The next generation of unicorns will emerge from the knowledge industry, leveraging intangible assets."

The IT industry has evolved from hardware manufacturing to software development, then to service-based platforms like Uber and knowledge hubs like Wikipedia. Now, as we step into the AI era, the focus shifts toward replicating human consciousness—a trend that's already reshaping startup valuations. 

According to Belozerov, the most valuable players will be those pioneering the AI-driven replication of human cognition.

Safe Superintelligence, a new startup launched by Ilya Sutskever, former chief scientist and co-founder of OpenAI, exemplifies this approach. The startup is focused on addressing the most important technical problem of our time—the risks posed by AI. 

Sutskever stated that SSI's first product will be safe superintelligence, and the company "will not do anything else" until that goal is achieved.

Catalyst for Creation and Destruction

Belozerov admits that AI is an unfinished product. "There are still risks that prevent it from being fully integrated into business processes," he says.

For example, many small and mid-sized banks in the U.S. have prohibited the use of applications like ChatGPT due to concerns about security, hallucinations, and biases inherent in AI models based on the data used to train them.

This means that AI investments typically take a while to pay off. Belozerov uses self-driving cars as an example, highlighting the complexity of the technology.

"Consider how many self-driving cars rely on Light Detection and Ranging (LIDAR), a remote sensing method," he said. "These investments take time. This is why it's crucial to create a diversified portfolio to mitigate the uncertainties associated with AI."

According to Belozerov, one of the risks is AI developing new behavioral strategies.

A few years ago, a story about a gamer who set up a bot-vs-bot match on his Quake III Arena server went viral on Twitter. After forgetting about the match for four years, the gamer logged back in to find the bots eerily standing still, not attacking despite the server running smoothly. 

"Not sure if it's true, but humans often rely on emotions and personal experiences in decision-making, while superintelligence focuses on optimal strategies," Belozerov said. 

In a recent experiment, researchers found that AI could intensify conflicts. The study created a simulated game with fictional countries of varying military strength, using models like GPT-4, GPT-3.5, Claude 2.0, Llama-2-Chat, and GPT-4-Base.

"The results show that all models could potentially destroy the world, with GPT-3.5 being the most aggressive," Belozerov said. "While the exact cause remains uncertain, biases in the training data might be a contributing factor."

Investing in Trends and Leaders

One major growth area Belozerov highlights is the intersection of AI and cybersecurity. According to a new report by MarketsandMarkets, the global generative AI cybersecurity market is expected to grow from an estimated USD 7.1 billion in 2024 to USD 40.1 billion by 2030.

"AI cybersecurity will see significant development as an industry," he notes. "AI is used to prevent different types of attacks and security threats. At the same time, new technologies are being developed to keep AI itself secure."

The recent global IT blackout caused by a software update from a cyber firm underscores the critical importance of this sector. 

In July, widespread outages affected banks, airlines, television networks, and health systems globally, demonstrating the vulnerability of essential services. Thousands of flights and train services were canceled, including over 1,800 in the U.S., leading to significant disruptions.

Belozerov advises that while AI is a significant trend, the same investment strategy applies to any sector. "Investing in a broader trend and then focusing on the leaders within that trend can yield promising returns," he concludes.

According to Belozerov, emerging AI trends involve a growing distinction between public neural networks, like LLMs, and private, industry-specific models tailored for applications such as healthcare. 

"Companies are increasingly developing bespoke AI systems to address unique needs, such as creating specialized analyzers for medical imaging," Belozerov said. "Additionally, a significant trend is the replacement of human roles by AI, such as the shift from human traders to AI-driven robots in the stock market."