AlphaWire

newswire

AMD says new product performance is expected to rival Nvidia Open AI and other customers promise mass adoption

2026-07-25·newswire-us-stock-204219
AMD says new product performance is expected to rival Nvidia Open AI and other customers promise mass adoption.

(AMD) has released a series of new products and say they will outperform competitors similar products, in order to achieve rapid growth in Further expand its share in the (AI) computing power market. At a press conference held on Thursday (July 23), AMD launched a new generation of server processors, upgraded AI accelerators and a new server rack design.

AMD said that these products are a major upgrade to the existing product line and can compete with or even have advantages over competitors' products in terms of performance. The conference was held in San Francisco.

Su Zifeng also invited executives from a number of technology companies to the stage, many of whom promised to use AMD products, including the two major AI model developers Anthropic and OpenAI. At the press conference, AMD highlighted its new generation Helios server rack.

This design will be manufactured and sold by server manufacturers themselves, and can integrate up to 72 MI400 series AI accelerators. Among them, the MI455 model has achieved significant improvements in terms of memory bandwidth, inference speed and cost efficiency compared with the previous generation products.

OpenAI executive Sachin Katti said that as the company continues to expand its AI infrastructure, it expects "large-scale deployment" of Helios servers.

"If you haven't heard yet, I need more computing power, and the sooner the better." Su Zifeng Sachin Katti At the same time, AMD has also upgraded its Epyc server processor product line and launched new products using the new Venice architecture. AMD says this product will continue to lead in performance Upcoming Vera server CPU.

In addition, AMD also announced a cooperation with chip company Cerebras Systems. The two parties will jointly launch a solution combining Helios servers and Cerebras servers for customers with extremely high requirements for AI response speed.

According to the plan, AMD's chips are responsible for parsing user queries, while Cerebras' chips with ultra-large on-chip memory are responsible for quickly generating answers. The new platform will be deployed first on Cerebras’ own , and then provide commercial services to other AI companies.

Cerebras CEO Andrew Feldman said the program will It is the first to go public before integrating similar technology launched from its recent acquisition of Groq-related assets. He said the company has been developing the joint product for months and is confident of its leading edge. "We're going to be first to market and we're confident about it.

This is a market for customers who place a premium on responsiveness." Andrew Feldman Su Zifeng AMD expects the market for AI accelerators (chips used to train and run AI models) to reach $1.4 trillion by 2030. At the same time, AMD’s traditional The CPU market size will reach 220 billion US dollars.

By 2030, the total market size in which it competes will reach US$2 trillion. Currently, NVIDIA relies on The technology has dominated the global wave of AI data center construction and created hundreds of billions of dollars in revenue. But AMD is becoming more and more confident and is no longer just a "long-distance chaser" of Nvidia.

The same thing is that both companies believe that customer demand for AI computing power is still extremely strong, so they are racing to launch more powerful chip products.

Under the leadership of Su Zifeng, AMD has achieved rapid growth in recent years, and Wall Street also expects its growth rate to further accelerate, but it is still difficult to challenge Nvidia.

Although AMD is widely regarded as one of the strongest competitors, Nvidia is still growing at a faster rate, and its single quarter revenue even exceeds AMD's full year revenue.

#Stocks #Nvidia #AMD #AI #Semiconductors

Full text

AMD says new product performance is expected to rival Nvidia Open AI and other customers promise mass adoption

[AMD claims that the performance of its new products is expected to be comparable to that of Nvidia, and customers such as OpenAI are committed to large-scale adoption] AMD has released a series of new products for data centers and said that the performance of these products will surpass similar products of rival Nvidia in order to further expand its share in the fast-growing artificial intelligence (AI) computing power market.

(AMD) has released a series of new products and say they will outperform competitors similar products, in order to achieve rapid growth in Further expand its share in the (AI) computing power market. At a press conference held on Thursday (July 23), AMD launched a new generation of server processors, upgraded AI accelerators and a new server rack design. AMD said that these products are a major upgrade to the existing product line and can compete with or even have advantages over competitors' products in terms of performance. The conference was held in San Francisco. Su Zifeng also invited executives from a number of technology companies to the stage, many of whom promised to use AMD products, including the two major AI model developers Anthropic and OpenAI. At the press conference, AMD highlighted its new generation Helios server rack. This design will be manufactured and sold by server manufacturers themselves, and can integrate up to 72 MI400 series AI accelerators. Among them, the MI455 model has achieved significant improvements in terms of memory bandwidth, inference speed and cost efficiency compared with the previous generation products. OpenAI executive Sachin Katti said that as the company continues to expand its AI infrastructure, it expects "large-scale deployment" of Helios servers. "If you haven't heard yet, I need more computing power, and the sooner the better." Su Zifeng Sachin Katti At the same time, AMD has also upgraded its Epyc server processor product line and launched new products using the new Venice architecture. AMD says this product will continue to lead in performance Upcoming Vera server CPU. In addition, AMD also announced a cooperation with chip company Cerebras Systems. The two parties will jointly launch a solution combining Helios servers and Cerebras servers for customers with extremely high requirements for AI response speed. According to the plan, AMD's chips are responsible for parsing user queries, while Cerebras' chips with ultra-large on-chip memory are responsible for quickly generating answers. The new platform will be deployed first on Cerebras’ own , and then provide commercial services to other AI companies. Cerebras CEO Andrew Feldman said the program will It is the first to go public before integrating similar technology launched from its recent acquisition of Groq-related assets. He said the company has been developing the joint product for months and is confident of its leading edge. "We're going to be first to market and we're confident about it. This is a market for customers who place a premium on responsiveness." Andrew Feldman Su Zifeng AMD expects the market for AI accelerators (chips used to train and run AI models) to reach $1.4 trillion by 2030. At the same time, AMD’s traditional The CPU market size will reach 220 billion US dollars. By 2030, the total market size in which it competes will reach US$2 trillion. Currently, NVIDIA relies on The technology has dominated the global wave of AI data center construction and created hundreds of billions of dollars in revenue. But AMD is becoming more and more confident and is no longer just a "long-distance chaser" of Nvidia. The same thing is that both companies believe that customer demand for AI computing power is still extremely strong, so they are racing to launch more powerful chip products. Under the leadership of Su Zifeng, AMD has achieved rapid growth in recent years, and Wall Street also expects its growth rate to further accelerate, but it is still difficult to challenge Nvidia. Although AMD is widely regarded as one of the strongest competitors, Nvidia is still growing at a faster rate, and its single quarter revenue even exceeds AMD's full year revenue.

← Back to archive