In the current era of rapid AI development, computing power has become one of the key factors driving technological breakthroughs in AI. On November 3, OpenAI and Amazon Web Services (AWS) officially signed a $38 billion cloud computing service procurement agreement. According to the agreement, OpenAI will immediately begin using the computing power resources provided by AWS, initially utilizing hundreds of thousands of NVIDIA's latest GB200/GB300 series GPUs in the United States, with the capability to scale up to tens of millions of CPUs. These resources will support OpenAI's core large-model training, inference deployment, and product innovation needs, covering a wide range of tasks from real-time responses of ChatGPT to the training of next-generation models.

From Microsoft to AWS: Strategic Adjustment and Diversified Cooperation

For a long time, Microsoft has been OpenAI's closest strategic partner and the main infrastructure provider. Since 2019, Microsoft has invested a total of $13 billion in OpenAI and provided strong computing power support through its Azure cloud platform. However, as AI market competition intensifies and tech giants like Google and Meta accelerate the development of their own large-scale models, OpenAI urgently needs to reduce its dependence on a single supplier to ensure its technological direction and business decisions are not constrained.

This cooperation with AWS is a key step in OpenAI's "de-Microsoft" strategy. By migrating part of its core workload to AWS, OpenAI can not only obtain a powerful alternative computing power source but also use market competition mechanisms to create a balance of power between Microsoft and Amazon. This will give OpenAI more initiative in price negotiations, technical support, and resource prioritization.

Opportunities and Challenges for AWS

For AWS, this $38 billion order is not only a significant breakthrough in the AI computing power market but also a key move to strengthen its competitiveness in the cloud computing field. AWS Chief Executive Officer Matt Garman said that AWS's infrastructure security and performance can "fully support OpenAI's ambitions." The breadth and instant availability of its optimized computing power fully demonstrate AWS's unique advantages in supporting OpenAI's large-scale AI workloads. However, AWS also faces pressure from competitors such as Microsoft and Google Cloud. The cloud businesses of Microsoft and Google are growing faster, with their cloud service revenues increasing year-on-year by 40% and 34%, respectively, while AWS's departmental revenue increased by more than 20%, exceeding analysts' expectations but still needing to strive to catch up.

Future Outlook

This cooperation will strengthen the computing power ecosystem and promote AI development. OpenAI co-founder and CEO Sam Altman emphasized, "Expanding cutting-edge AI requires massive and reliable computing power resources. Our cooperation with AWS will improve the computing ecosystem and lay the foundation for next-generation intelligent technologies." Within the next seven years, OpenAI can further expand its computing power scale through AWS according to its needs. However, the specific plans after 2026 have not yet been finally determined. In addition, OpenAI's basic models (including the so-called "open-weight" versions) are now available to customers through AWS's managed AI service Bedrock, which will further promote the popularization and application of AI technology.

ConevoSemiconductor Chip Distributor

Conevo, an integrated circuit chip distributor, is committed to providing efficient, reliable and diverse IC solutions for shortages. The Conevo chip product category is rich, covering many popular IC types such as Logic devices, Analog ics, Microcontrollers (MCUs), Power Management chips, and Memory (such as NOR Flash, NAND Flash). Here are some popular chip models:

● 74LVC1G14GW-Q100: A high-performance logic device suitable for signal processing and control in a variety of electronic devices.

● LTC6957IDD-3#PBF: A high-precision analog IC, often used in applications that require precise clock signals.

● DSPIC30F6014A-30I/PF: A powerful microcontroller suitable for motor control and digital signal processing.

Website: www.conevoelec.com

Email: info@conevoelec.com