① Chairman Liu Qingfeng said that Q1 revenue was relatively low throughout the year, and the continued increase in R&D investment was the main reason for the increase in losses; ② The company plans to release the Spark Model in June this year to be able to benchmark the GPT-4 Turbo level, but the base capacity will still catch up dynamically in the long run.
Financial Services Association, April 23 (Reporter Liu Mengran) HKUST iFLYTEK (002230.SZ), which lost 300 million dollars in Q1 net profit and a decline in gross profit this year, was asked by many investors at the 2023 annual results briefing today about the reasons for the low Q1 gross margin and their expectations for future trends. Liu Qingfeng, chairman of the company, explained during the exchange that losses were due to a relatively low share of Q1 revenue throughout the year, and due to continued increase in R&D investment. There was no decline in gross margin due to intense market competition, and the annual profit is still better than last year.
According to financial reports, iFLYTEK achieved revenue of 19.65 billion yuan in 2023, an increase of 4.41% year on year, and net profit attributable to the parent company was 657 million yuan, an increase of 17.12% year on year. In the first quarter of this year, the company achieved revenue of 3,646 billion yuan, an increase of 26.27% over the previous year, but net profit loss attributable to shareholders of listed companies was 300 million yuan, compared to a loss of 578.953 million yuan in the same period last year.
Regarding the Q1 net profit loss, Jiang Tao, the company's senior vice president, said that the company is investing heavily in the Starfire Big Model and will not lag behind in the development of GM's big model base. In 2023, the company invested 3,839 billion yuan in R&D, accounting for 19.53% of revenue, an increase of 482 million yuan over the previous year. Of these, the total R&D investment of the Spark Big Model exceeded 2 billion yuan. While increasing R&D investment, the Spark Model was actively marketed and promoted.
Liu Qingfeng further explained that the year-on-year increase in Q1 losses was also due to the fact that G-side and B-side revenue accounted for about 15% of Q1 revenue in previous years, but fixed amortization of R&D investment, including increasing R&D personnel and purchasing hardware equipment, was the same every quarter. The revenue share was relatively low, but R&D spending increased, leading to losses during the quarter. He said that as the share of C-side business increases, the share of Q1 revenue is expected to gradually increase to 20%.
The “internal volume” of the domestic big model continues, and iFLYTEK increased its investment in the development of the cognitive big model, and the impact on current profits has already been reflected. According to industry analysts, the commercialization model of the domestic AI model is still being explored. The prerequisite for achieving profit is to restructure user application scenarios on the C-side and use AI technology on the B-side to empower the industry at multiple levels.
Regarding the Q1 gross margin issue that investors are concerned about, Liu Qingfeng said that the first is to increase the adjustment of the three-block business to the C-side. Learning machines will continue to maintain 99% revenue growth, and gross profit will increase by 104%, but compared with conventional software-only systems such as education solutions, gross margin will be slightly lower; in addition to voice interaction technology solutions, the automotive business has added automotive sound effects projects, leading to a decrease in gross margin.
At the same time, in the G-end education business, some local authorities will require that smart hardware be completed before software is installed, so the company needs to confirm revenue according to the progress of the implementation project. Liu Qingfeng said that in the future, the overall gross profit of the education business will remain above 50% to guarantee continuous back-office R&D capabilities and provide operation and maintenance services.
According to Jiang Tao, senior vice president of the company, the C-side share of the company's revenue structure continued to rise in 2023. “Open Platform and Consumer Business” revenue was 6.19 billion yuan, accounting for 31.47%. In addition, G-side and B-side revenue were 32% and 26% respectively.
On the business side, Liu Qingfeng revealed at the performance conference that it is planned that the version of the Spark Big Model released in June of this year can match the GPT-4 Turbo level. However, he also mentioned that if the GPT-5 version is released, then the gap will be widened again, so it will also be a process of dynamic catch-up in terms of base capacity.
However, the operating cash flow of the Q1 company this year increased by 284 million compared to last year's Q1, and sales repayment of 4.36 billion yuan, an increase of 18.79% over the previous year, exceeding the Q1 revenue scale.