Under the wave of artificial intelligence, iFLYTEK continues to strengthen AI empowerment, and the trend of "sparks can set fire to the plain" is gradually emerging.
Recently, iFLYTEK released its 2023 semi-annual report. According to the financial report, in the first half of 2023, iFLYTEK achieved an operating income of 7.842 billion yuan and a net profit attributable to shareholders of listed companies of 74 million yuan. Specifically, the second quarter significantly reversed the downward trend in the first quarter, achieving a sequential increase of more than 60% in revenue and gross profit.
The main reason why the second quarter had such a bright financial performance is the success of the Spark model in the C-end market. The capital market is also optimistic about iFLYTEK, and the share price of iFLYTEK has risen by nearly 98% this year, and the number of shareholders of the company as of June 30, 2023 is 234,600, an increase of 12.40%, far higher than the industry average of 38,700. In addition, a total of 28 institutions have rated iFLYTEK in the past 90 days, including 23 buy ratings and 5 overweight ratings.
However, it is worth noting that although the "Spark Cognitive Big Model" has helped iFLYTEK open up the space for performance growth, the market competition for AI large models is becoming more and more fierce, how can iFLYTEK, as an AI veteran, maintain its competitive advantage?
C-end revenue hits record high Spark cognitive big model accelerates the release of product power
From 2022 to the first quarter of 2023, due to supply chain adjustments and large-scale model research, iFLYTEK continued to increase cost expenditure, and new investment also affected performance to a certain extent. The revenue in the first quarter of 2023 was 2.888 billion yuan, a year-on-year decrease of 17.64%.
But from the second quarter, everything started to improve. According to the financial report, iFLYTEK's operating income in the second quarter was 4.954 billion yuan, a year-on-year increase of 9.68% and a month-on-month increase of 71.56%; gross profit was 1.938 billion yuan, a year-on-year increase of 7.80% and a month-on-month increase of 60.21%; In addition, net cash flow from completed operating activities was $140 million, an increase of $575 million over the same period last year.
For example, in the first half of 2023, the revenue of education, open platform and consumer, smart car, smart healthcare, smart finance and other businesses increased by 3.63%, 21.6%, 26.14%, 37.13%, and 22.19% year-on-year, respectively, but the key role is that the Spark cognitive big model brings significant business increments.
It is understood that on May 6, Xinghuo cognitive big model and its commercial application results were released simultaneously, such as large model + AI learning machine, large model + iFLYTEK hearing, etc. With the blessing of the large-scale marketing network laid over the years, the transformation of product power into sales is very rapid.
The financial report disclosed that iFLYTEK has set up a total of 32 regional marketing agencies (provinces + central state-owned enterprises) across the country, covering all provinces in China, and among more than 2,800 districts and counties in the country, iFLYTEK has effectively reached 85.7% of the main customer groups in the districts and counties. The GMV of C-side hardware thus reached an all-time high. According to the financial report, the GMV of iFLYTEK AI learning machine equipped with the Xunfei Xinghuo cognitive large model increased by 136% and 217% in May and June respectively, and the gross profit of sales in the first half of the year increased by 60% year-on-year, growing strongly and outperforming the industry market.
The good commercialization results of iFLYTEK's AI model are also related to its long-term product strategy of focusing on the C-end market.
At present, many AI large model manufacturers have focused their development on the B-end market, and have launched industry large models for vertical scenarios, such as Tencent's MaaS large-model one-stop service platform for several industries such as finance and government affairs, and Huawei's industry-oriented Pangu model has also been unveiled recently. However, such large-model products often consider how to improve the efficiency of management, organization, and business levels, which requires a long product incubation period.
The C-end market is different, based on the consumer's focus on experience, the product can solve a practical need, and the operation is simple and easy to use, and can get market feedback faster. For example, iFLYTEK AI learning machine, equipped with the iFLYTEK Spark cognitive model, the experience of Chinese and English composition tutoring, Chinese and English oral learning has been significantly improved, and the good product experience has quickly transformed into real sales.
The sales volume is gratifying, and the management is also full of confidence in the future, and disclosed the share repurchase plan in the financial report, reflecting the confidence in the company's long-term intrinsic value, and expects that on October 24 this year, iFLYTEK Xinghuo cognitive large model will achieve a universal model benchmark against ChatGPT, and achieve Chinese surpassing, English equivalence. However, at this stage, the development of AI large models still faces many challenges, as technology giants have increased investment, how should iFLYTEK build its own moat?
Giants accelerate the layout of AI big models How does iFLYTEK build a moat?
In the current situation, it is not easy to make a big model. On the one hand, the market continues to expand, and competition continues to intensify. At present, the world basically regards artificial intelligence as an important force in a new round of scientific and technological revolution and industrial transformation, it is predicted that the scale of the core industry of artificial intelligence in mainland China will exceed 1 trillion yuan in 2030, and the global artificial intelligence market will reach 16 trillion US dollars, and there is still great potential and space for development. Under the background of the overall improvement of the industry, manufacturers have accelerated the pace of large model layout.
On the other hand, the commercialization of large models is still in the early exploration stage and requires a lot of investment. Microsoft has invested more than $10 billion in OpenAI so far, and OpenAI CEO Sam Altman admitted that OpenAI may need to try to raise as much as $100 billion in the next few years to develop sufficiently advanced general-purpose AI while keeping the company running.
Domestic AI large model companies, whether they are leading companies such as iFLYTEK or cutting-edge companies such as Yuncong Technology, have shown a general upward trend in R&D expenses. According to the financial report, iFLYTEK's R&D investment increased from 2.936 billion yuan in 2021 to 3.355 billion yuan in 2022, a year-on-year increase of 14.28%; The number of R&D personnel increased from 8367 to 9281, a year-on-year increase of 10.92%, and the proportion of employees increased to 61.68%. In the first half of 2023, although facing the pressure of revenue and profit, R&D investment also further increased to 1.656 billion yuan, a year-on-year increase of 3.57%. And the competition for technology investment is far from over, which tests the cash flow of enterprises.
From this point of view, although iFLYTEK has achieved certain scale results in a number of C-end market segments, for example, the financial report shows that iFLYTEK translation machine ranks first in the domestic market share, sales in the first half of 2023 increased by 600% year-on-year, and the shipment of new front-mounted intelligent products in the automotive field is nearly 4.1 million sets, with a cumulative shipment of more than 50 million sets, and the financial technology uses AI + big data to empower inclusive insurance, covering 53 cities and a cumulative number of insured users of 15.45 million, but the challenges in the future cannot be underestimated.
So, under this situation, what will iFLYTEK rely on to achieve further breakthroughs in the commercial application of large models?
On the whole, AI large model companies are iterating technology while laying scenarios to seek commercial breakthroughs, and in such a rally, self-hematopoietic ability is crucial. From the financial report, iFLYTEK has achieved the transformation of product sales at the general model level, and its strong self-hematopoietic ability can help it stabilize its position in the "thousand model war", and behind this requires iFLYTEK to continuously develop rich application scenarios in cognitive intelligence with a solid foundation of AI core technology research and development capabilities.
Specifically, iFLYTEK's R&D direction is mainly based on source innovation, not simply generative AI R&D or product development connected to other B-side AI ecosystems.
The biggest advantage of mastering source technology is that the core technology research and product development platform is independent and controllable, which can build a development ecosystem based on consumer needs. According to the financial report, iFLYTEK Open Platform has opened 587 AI capabilities and scenario solutions, gathered 4.974 million developer teams, reached a total of 1.725 million applications, linked more than 5 million ecological partners, and the total number of developers on the international website of iFLYTEK Open Platform has reached 138,000.
Of course, in addition to iFLYTEK, Huawei, Alibaba Cloud and other enterprises are also vigorously building a developer ecosystem, Huawei invested 1.94 billion yuan in 2022 through developer plans such as Horizon Cloud Innovation Program, CrowdIntelligence Program, and Star Wealth Project to complete the empowerment of more than 3.65 million developers; Alibaba Cloud's Modai community also gathered more than 2 million AI developers as of the end of July 2023. However, in terms of scale and number of transformation results, iFLYTEK is in the lead. In addition, from the perspective of the product series of Jingdong Mall, only iFLYTEK currently has a relatively complete AI hardware product system, and Huawei and Alibaba Cloud have fewer product lines on the C-end.
The development of the C-end is faster than that of the B-side, and with the complete product series on the C-side, iFLYTEK can gather more developers. The ecology can be regarded as the door to the commercialization of the AI big model, and the enterprises that take the lead in building and forming can have more say in the next stage of competition, and based on the subsequent improvement of the ecology, iFLYTEK is expected to explore greater growth space.
The financial report revealed that in the second half of 2023, iFLYTEK Spark Cognitive Model will continue to break through and upgrade, and on August 15, iFLYTEK Spark Cognitive Model has also released a V2.0 upgraded version, showing the capabilities and corresponding achievement applications of multimodal understanding (including visual question answering, description, reasoning, etc.) and multimodal generation (text and image generation, controllable speech synthesis). After that, the sales performance of C-end intelligent hardware products may further verify the logic of iFLYTEK to develop an ecosystem to enhance product transformation.
Author: Good blue is not working
Source: Songuo Finance