After climbing to the top of the world's highest market capitalization company last month, Nvidia also faced "giant troubles".
According to Reuters, direct sources have revealed that French antitrust regulators will charge Nvidia on suspicion of anticompetitive behavior, marking the first time the graphics card giant has faced official antitrust charges.
As early as September last year, the French antitrust agency launched a surprise investigation into a company suspected of engaging in anti competitive behavior in the graphics card industry. The Wall Street Journal later published a report confirming that this company is Nvidia.
Last Friday, the French antitrust agency released another report on competition in generative artificial intelligence, which specifically mentioned the current risk of chip suppliers abusing AI technology.
According to the latest report from Reuters, French regulatory authorities are mainly concerned about Nvidia's monopoly on the market, focusing on two things: first, the graphics card industry's reliance on Nvidia's CUDA chip programming software, and second, Nvidia's latest investment in start-up cloud computing service companies such as CoreWeave, which focus on artificial intelligence.
Currently, both the French authorities and Nvidia have refused to comment.
In the past, antitrust investigations were considered special treatment for giants. Microsoft, Google, Amazon, and Apple are all frequent customers of antitrust investigations in the United States and the European Union. Nowadays, Nvidia's global influence cannot be ignored.
Last month, Nvidia's stock price saw a significant increase, with a market value exceeding $3 trillion, surpassing Apple and Microsoft to become the world's highest value company, but then sliding again. At present, the latest market value of the company is 3.06 trillion US dollars, ranking third behind Microsoft and Apple (with market values of 3.39 trillion US dollars and 3.32 trillion US dollars respectively).
After ChatGPT's explosive popularity drove a wave of generative artificial intelligence technology, Nvidia became the "biggest winner" in this wave due to its high-end GPU graphics card, which is essential for training AI models. Its A100, H100 and other GPUs have been fiercely sought after worldwide, making it difficult to obtain a single card.
According to international research firm Techinsights, in terms of data center GPU sales revenue, Nvidia accounted for 98% of the entire market share in 2023. Gartner once estimated that even from a global perspective in the AI chip industry, Nvidia's market share may have reached 90%.
A dominant market share makes Nvidia's influence crucial. Goldman Sachs once wrote a report pointing out that the shortage of AI chip supply last year was mainly due to the shortage of Nvidia H100 chips. Currently, Nvidia's new generation H200 and B200GPUs will also begin large-scale shipments in the second quarter of this year. According to NVIDIA CEO Huang Renxun, the company is racing against supply and delivery every day, and the products are still in short supply.
In addition to its large market share, Nvidia's other secret weapon, CUDA, is the focus of this official investigation into its alleged monopoly. CUDA is a software development platform provided by Nvidia for its own GPUs, including drivers, compilers, math libraries, and other software tools. Since its release in 2006, Nvidia has established a complete ecosystem around CUDA. The daily development, maintenance, and functional upgrades of GPUs cannot be separated from this platform. According to statistics, over 4 million developers worldwide rely on Nvidia's CUDA. Therefore, in the eyes of many industry professionals, CUDA is Nvidia's true "moat",
In the first half of this year, when Nvidia updated its CUDA version, a new clause also confirmed the importance of its CUDA ecosystem. The new clause clearly states that "users are not allowed to reverse engineer, decompile, or disassemble any results generated using the SDK, and translate them on non NVIDIA platforms.". At that time, market analysts believed that Nvidia's move was intended to restrict other GPU manufacturers and prevent competitors from easily bypassing its established CUDA ecosystem through technological means.
To compete against Nvidia, competitors are actively forming an alliance.
In April of this year, Intel, Qualcomm, and semiconductor manufacturers such as Samsung and Arm led the establishment of the Unified Acceleration Foundation, which Google later announced to join. This foundation is known to the outside world as& quot; The goal of the Anti CUDA Alliance is to start from the software level and build an open-source project that can support multiple AI accelerator chips, in order to reduce dependence on the Nvidia CUDA platform.
In addition to CUDA, Nvidia's other moats also face challenges from competitors. Previously, according to multiple media reports, companies such as AMD, Intel, Google, Microsoft, Broadcom, Cisco, and Meta have formed the UALink (Ultra Accelerator Link) alliance, aimed at countering NVLink, another hardware interconnect technology of Nvidia.
According to French antitrust regulations, companies judged to be monopolistic may face a fine of 10% of their annual revenue. The largest fine imposed by the agency since 2011 was 1.24 billion euros (approximately 1.33 billion US dollars, with Apple's revenue of 274.5 billion US dollars) imposed on the company in 2020. But there are also companies surveyed who choose to make concessions to avoid fines, and it is unclear whether Nvidia will choose to make concessions.
As of the time of publication, Nvidia's pre market stock price fell by 1.26%.