Musk continues to make significant strides in the field of AI.
On February 23, Musk announced that the early beta version of Grok Voice Mode is now available on the Grok app. Grok 3’s Voice Mode allows users to communicate with artificial intelligence through natural language, facilitating more intuitive and efficient information acquisition and interaction.
According to photos posted by Musk earlier, the Grok app has topped the list of free apps in the U.S. App Store, surpassing OpenAI’s ChatGPT app. The xAI team revealed that they have begun the next stage of AI cluster construction, which will be about five times more powerful than the current cluster.
The Intensifying Global “AI War”
Domestically, with the release and growing popularity of the DeepSeek-R1 model, AI application models are undergoing significant transformations. Following mainstream domestic mobile phone manufacturers such as Huawei and Honor, Xiaomi has also officially integrated DeepSeek-R1. Additionally, major domestic cloud computing giants—including Huawei Cloud, Tianyi Cloud, Tencent Cloud, Alibaba Cloud, Volcano Engine, China Unicom Cloud, China Mobile Cloud, and Baidu Smart Cloud—have also integrated DeepSeek.
Musk’s Announcement
On February 23 (Beijing time), Musk announced on social platform X that the early beta version of Grok Voice Mode is now available on the Grok app. “Although this is an early beta version and may have some issues (which we will quickly resolve), it is still great,” he said.
While Musk acknowledged that Grok’s Voice Mode is still in its early testing phase and may have some issues, he praised its performance. The launch of Grok’s Voice Mode means that users can now interact with Grok through voice, significantly enhancing user experience and expanding Grok’s application scenarios.
Grok 3’s Voice Mode allows users to engage in conversations with AI using natural language, enabling more intuitive and efficient information exchange. The mode offers two different voices (Ara and Grok), multiple personalities, customizable instructions, and a share button to record and share conversations.
On February 20, xAI announced that Grok 3 is now free to use (until server capacity is exceeded). Previously, access to Grok 3 required a subscription to X Premium+ or SuperGrok.
This move quickly created a buzz in the global AI community. Later, Musk posted an image of the Apple App Store’s free app rankings, showing that the Grok app had surpassed the ChatGPT app to claim the top spot.
Musk even claimed that Grok 3 will be used for SpaceX’s Mars mission calculations in the future and predicted that “a Nobel Prize-level breakthrough will be achieved within three years,” further fueling excitement about Grok 3’s potential.
On February 18, Musk and the xAI team officially launched Grok 3 in a live broadcast.
Based on the data presented during the live session, Grok 3 outperformed Google’s Gemini model, Anthropic’s Claude model, and OpenAI’s GPT-4o model in assessments related to mathematics, science, and programming.
Musk stated that after being trained with synthetic data, Grok 3 can analyze its mistakes through reflection, leading to improved logical consistency. The most notable feature of Grok 3 is the introduction of the “Chain of Thought” reasoning mechanism, which enables the AI to solve complex problems step by step, mimicking human cognitive processes.
“Grok 3 has over ten times the computational power of Grok 2,” Musk stated. “We are continuously improving the model every day.”
For the first time, the xAI team also revealed that they have started the next phase of AI cluster construction. “A better model than Grok 3 must excel in all aspects of deep learning science and engineering, but this is by no means an easy task,” the team explained during the live broadcast. “We have already started working on the next AI cluster, which will be about five times more powerful than the current one.”
The Rise of AI Applications
As the “AI war” escalates, industry insiders predict that 2025 will mark the explosion of AI terminal applications.
Recently, with the release and widespread adoption of the DeepSeek-R1 open-source model, AI development and application models are undergoing significant changes, drawing widespread attention both domestically and internationally.
Following Huawei and Honor’s announcement of DeepSeek-R1 integration, Xiaomi has also officially adopted DeepSeek-R1. On February 23, Xiaomi’s Super Xiaoai integrated DeepSeek-R1, which defaults to online search mode. Users can activate it by entering “Open Deep Thinking.”
Additionally, since the DeepSeek open-source model can be deployed on both public and private clouds, domestic cloud computing giants—including Huawei Cloud, Tianyi Cloud, Tencent Cloud, Alibaba Cloud, Volcano Engine, China Unicom Cloud, China Mobile Cloud, and Baidu Smart Cloud—have all integrated DeepSeek.
An Yun, Deputy Director of the Artificial Intelligence Research Institute of Saizhi Industry Research Institute, stated that domestic AI models have achieved breakthrough technological progress through innovations such as open-source strategies, low-cost and efficient reasoning, and reinforcement learning combined with a hybrid expert architecture (MoE). Open-source development has disrupted the technological monopoly of large enterprises and accelerated the widespread adoption of AI technology. Its cost-effective algorithm optimization model has reduced dependence on computational power while shifting the competitive landscape towards efficiency.
At the same time, domestic AI models have impacted various industries, from manufacturing—where they optimize production processes and quality control—to finance, where AI-driven risk assessment has significantly improved efficiency. In the medical sector, AI-assisted diagnosis is shortening treatment cycles. Additionally, AI-driven decision-making in government, education, and transportation is improving both efficiency and service quality.
Guojin Securities believes that if 2023 was the year of AI training and 2024 marked the beginning of AI reasoning, then 2025 will be the breakthrough year for AI terminal applications. Domestic AI models have entered a new stage of global development and application, reinforcing confidence in AI infrastructure and application advancements.
Minsheng Securities points out that the cloud industry will be the biggest beneficiary of DeepSeek. Cloud resources have become “hard currency,” and vendors that control computing resources will hold a significant advantage. As enterprises race to adopt DeepSeek, cloud vendors with ample computing power and broad user coverage are expected to see rapid growth.
Analysts predict that cloud computing, as the foundational infrastructure for AI models, will continue to benefit. Advances in AI computing efficiency will not only support traditional AI computing power providers but also extend to non-traditional large-scale cloud vendors. As leading cloud computing firms integrate DeepSeek, demand for cloud computing services is expected to rise significantly. For domestic cloud vendors, DeepSeek will accelerate enterprise cloud adoption and enhance profitability through economies of scale.
