AMD Challenges Nvidia with Next Generation of AI Infrastructure at Major San Francisco Event

AMD Unveils Major AI Hardware at Advancing AI Event

AMD launched its next generation of AI infrastructure at a two-day event in San Francisco, showcasing AI hardware designed to compete directly with Nvidia in the rapidly expanding data center chip sector 1

1

. At the Moscone West convention center, CEO Lisa Su headlined the company's most ambitious push yet into AI infrastructure spending, displaying its first-generation Helios server racks alongside cloud computing providers like Vultr and TensorWave 1

1

. The company also formally launched its EPYC Venice central processing unit for data centers, a critical component in AMD's strategy to capture market share in inference computing—the data crunching that occurs when users query chatbots like OpenAI's ChatGPT 1

1

.

Microsoft and Anthropic Deals Signal Growing Market Confidence

The chipmaker announced that Microsoft will deploy AMD's Helios rack-scale system on Azure to power advanced AI models, marking a significant milestone for the company 2

2

. Jefferies analysts noted that securing Microsoft as a customer represents "a major positive for the stock and reinforce confidence in AMD's ability to compete at the highest level without incentives" 2

2

. AMD also revealed plans to sell up to two gigawatts of its Instinct MI450 chips to Anthropic beginning in the first half of 2027, with the deal including an investment of as much as $5 billion in the Claude maker 1

1

. In October, AMD announced a multiyear deal with OpenAI that could bring in tens of billions of dollars in annual revenue while giving the ChatGPT creator the option to buy up to roughly 10% of the chipmaker 1

1

.

AMD vs Nvidia: The Battle Over Server CPUs and Throughput

The competition between AMD and Nvidia extends beyond traditional GPU performance to a fundamental debate about how AI infrastructure should be optimized. Bank of America analyst Vivek Arya explained that Nvidia measures success by how quickly a single AI agent completes its work, focusing on latency and "max single-threaded performance at scale" 4

4

. Nvidia's Vera CPU, combined with its Rubin GPU, aims to maximize work output per unit of electricity 1

1

. AMD takes a different approach, arguing that future AI systems will run thousands of agents simultaneously, shifting focus from speed to throughput 4

4

. According to Bank of America, AMD estimates its EPYC Venice platform could deliver about 3.3 times the rack-level throughput of Nvidia's Vera reference system 4

4

. The global server CPU market could reach $170 billion by 2030, four times its current size, making this architectural debate critical for both companies 3

3

.

Financial Performance and Market Opportunity Drive Investor Confidence

AMD reported first-quarter 2026 revenue of $10.3 billion, up 38% year-over-year, with its Data Center business reaching a record $5.8 billion, increasing by 57% 3

3

. Wells Fargo raised its price target on AMD to $615, suggesting the company's data center GPU revenue could reach $40.6 billion by 2027, with some investors expecting it could exceed $50 billion 3

3

. BNP Paribas raised AMD's price target to $600 from $460, citing higher valuation multiples as investors place a greater premium on companies positioned to benefit from the next phase of AI infrastructure spending 5

5

. AMD stock has risen roughly 140% this year, making it one of the best-performing S&P 500 stocks of 2026 2

2

. The firm believes the next AI spending wave will broaden across custom processors, Agentic AI CPUs, networking switch ASICs, high-speed optical transceivers, and memory 5

5

. For investors watching the AI chip market, the question isn't just about GPU performance anymore—it's whether hyperscalers will prioritize latency or concurrency as they build out massive AI infrastructure deployments 4

4

.