Artificial intelligence startup SSI (Safe Superintelligence), led by former OpenAI co-founder Ilya Sutskever, is currently in negotiations for a new round of funding, with the company's valuation expected to reach $20 billion, quadrupling its valuation from September last year. Although SSI has not yet generated any revenue, this enormous valuation reflects investors' confidence in its potential to develop transformative AI technologies.
Unlike other AI companies, SSI focuses on developing "safe superintelligence" rather than launching commercial products like ChatGPT first. The company is currently in a pure research phase, aiming to achieve artificial superintelligence (ASI). This ASI goal may also be one of the reasons Sutskever left OpenAI in May 2024, as his vision was not aligned with the increasingly commercial direction of OpenAI CEO Sam Altman.
According to Reuters, funding discussions are still in the early stages, and specific terms may change. The strong interest from investors largely stems from Sutskever's deep background in AI research. At the NeurIPS AI conference last December, Sutskever discussed what he calls the "data peak" problem, which refers to the limited data available for AI training, while computational power continues to grow. He compared training data to fossil fuels, suggesting that it will eventually be depleted.
To address this issue, Sutskever believes that potential future solutions include AI agents with independent thinking and reasoning abilities, synthetic data generation, and enhanced computational power during reasoning processes. Other large AI labs, such as OpenAI, Anthropic, Google, and DeepSeek, are also exploring similar solutions. Last November, Sutskever stated that the AI industry has entered a new "discovery era," where scaling remains important, but he emphasized the need to "scale the right things."
Key Points:
💰 SSI's valuation is expected to reach $20 billion, despite having no revenue yet.
🔬 SSI focuses on developing safe superintelligence, without prioritizing commercial products.
📊 Sutskever's "data peak" problem prompts the search for new solutions.
