iFlytek: The first-quarter loss expanded fourfold, how much money does the AI large model burn?

Since the release of ChatGPT by OpenAI, artificial intelligence has entered the era of large models. OpenAI has once again launched the DALL·E video model Sora, and with the release of the powerful Claude3 model by Anthropic, the race for large models continues to heat up. Among them, iFlytek, as a domestic AI star enterprise, claims that the release of the iFlytek Xinghuo cognitive large model has significantly promoted the growth of AI intelligent hardware sales. It also claims that in the first half of 2024, iFlytek Xinghuo will achieve a benchmark comparable to GPT-4.

In the first quarter of 2024, iFlytek's revenue reached 3.646 billion yuan, an increase of 26.27% year-on-year. However, despite the increased revenue, there was no increase in profit. This is mainly due to actively seizing the historical new opportunities of general artificial intelligence, and firm investment in areas such as general artificial intelligence cognitive large models. This resulted in a loss of 300 million yuan in Q1 2024, compared to a loss of 57.89 million yuan in the same period in 2023, an increase of 4.19 times.

In the first quarter of 2024, iFlytek added approximately 300 million yuan in new investment in large model research and development, core technology autonomy and controllability, and industrial chain controllability, as well as the expansion of the large model industry landing. Among them, R&D expenses were 842 million yuan, an increase of 126 million yuan compared to the same period last year. Although the above investment has affected short-term operating performance, it has further solidified the foundation for continuously consolidating iFlytek's position in the national artificial intelligence team industry.

On January 30, 2024, the first iFlytek Starlight V3.5 based on the national industrialized computing power platform was officially upgraded and released. Its language understanding, mathematical ability, and voice interaction capabilities surpass GPT-4 Turbo, with a code reaching 96% of GPT-4 Turbo and multimodal understanding reaching 91% of GPT-4V. In addition, the iFlytek Starlight voice large model was first released. In terms of multilingual speech recognition, the effects of the first batch of 37 mainstream languages surpassed OpenAI Whisper V3.

As a practitioner of applying artificial intelligence to the ground, iFlytek has been deeply cultivating the artificial intelligence track for more than twenty years. In the continuous and increasingly favorable industrial development environment of artificial intelligence, focusing on grasping the industrial opportunities of the artificial intelligence era in the long run, it has achieved a benign interaction between technological innovation and industrial application, with results emerging in the continuously expanding application scale.

At the same time, since the release of iFlytek Starlight until the end of 2023, the iFlytek open platform has added 1.676 million developer teams, an increase of 229% year-on-year, including over 350,000 large model developers. It has also attracted 27,000 assistant developers, developed 49,000 assistant applications, and is collaborating with more than 100,000 corporate customers to experience innovative applications using Starlight, accelerating the pace of industry empowerment. It has also jointly released industry large models with leading companies in 12 industries including finance, automotive, operators, and industry, fully integrating the capabilities of large models with various industrial fields, empowering various industries' digital transformation.

Finally, generative AI has taken the stage of the times, further promoting the development of artificial intelligence, reshaping various industries, and creating tremendous commercial value. Goldman Sachs predicts that generative AI based on large models will drive global GDP growth by 7% in the next decade, equivalent to nearly $7 trillion. According to Gartner's Top 10 Strategic Technology Trends for 2024, by 2026, over 80% of enterprises will be using generative AI.

Various sectors are committed to harnessing the transformative power of generative AI, and enterprises are seizing the opportunity to accelerate towards intelligence upgrades, using generative AI to transform their overall operations. IDC previously predicted that by 2027, global investment in generative AI solutions by enterprises will reach $143 billion.

Overall, with the breakthrough of large model technology, artificial intelligence is gradually entering the stage of general artificial intelligence development, which will further accelerate the promotion and application of digital and intelligent technologies globally, indicating that companies deeply cultivating the artificial intelligence track will usher in greater opportunities.