AMD CEO Lisa Su predicts five billion daily AI users by 2030

Company and market positioning

Advanced Micro Devices (AMD) has long positioned itself as a supplier of high‑performance CPUs and GPUs for data‑center workloads, gaming, and professional graphics. In recent years the company has pivoted to emphasize its role in the AI stack, offering GPU accelerators such as the Instinct series and a suite of software tools that integrate with popular frameworks like PyTorch and TensorFlow. The Shanghai AI Developer Day was a clear signal that AMD sees the next wave of growth coming from AI‑driven compute, not just traditional graphics or server markets.

The problem AMD is trying to solve

AI workloads are increasingly demanding more arithmetic throughput, memory bandwidth, and energy efficiency than conventional applications. Data‑center operators, cloud providers, and enterprises are wrestling with a looming compute gap: the hardware needed to run large language models, generative image tools, and real‑time inference at scale simply does not exist in sufficient quantity today. AMD argues that its heterogeneous architecture—combining Zen‑based CPUs with Instinct GPUs and a unified software stack—can close that gap more cost‑effectively than competing solutions.

Lisa Su’s forecast and its implications

During her keynote, Lisa Su said she expects around five billion people to interact with AI on a daily basis by 2030. That figure translates to roughly 63 % of the projected global population, suggesting AI will become as ubiquitous as the internet is today. Su emphasized that this is not hype; she described AI as a foundational technology that will drive sustained demand for compute infrastructure.

She also warned that overall computing requirements could increase a hundred‑fold, pushing the industry toward the 10 yottaFLOPS (10^24 floating‑point operations per second) scale. To put that in perspective, today’s combined global supercomputing capacity is measured in the low exaFLOPS range. A hundred‑fold jump would require a radical shift in hardware design, power delivery, and cooling solutions—areas where AMD is already investing heavily.

Traction in Greater China

Su highlighted AMD’s growing engineering presence in the region: more than 4,000 engineers across R&D centers in Greater China, with dedicated AI hubs in Beijing, Shanghai, Shenzhen, and Taipei. These hubs focus on custom silicon, driver development, and ecosystem partnerships with local cloud providers and chipset manufacturers. The regional emphasis reflects both the talent pool and the rapid adoption of AI services in China’s enterprise and consumer markets.

What this means for investors and partners

• Infrastructure spend: Cloud providers will need to refresh a large portion of their fleets in the next five years to keep up with AI demand, creating a steady revenue stream for GPU and CPU vendors.
• Software ecosystem: AMD’s open‑source initiatives, such as the ROCm platform, aim to lower the barrier for developers to port workloads onto Instinct GPUs. A broader software base could accelerate adoption and lock‑in customers.
• Supply‑chain considerations: Scaling to yottaFLOPS levels will stress semiconductor fabs, packaging, and testing facilities. AMD’s partnership with TSMC and its own fab‑less model give it flexibility, but the company will need to manage lead times carefully.

Context and next steps

The prediction aligns with other industry signals: OpenAI’s ChatGPT usage already exceeds a hundred million daily active users, and generative AI tools are being embedded in productivity suites, social media, and even consumer appliances. If the five‑billion figure holds, AI will move from a specialized tool to a daily utility, much like email or video streaming.

AMD’s roadmap, announced at the same event, includes the upcoming Instinct MI300X accelerator, which promises double‑digit performance gains over the current generation. Combined with new Zen 4‑based server CPUs, the company is betting that a tightly coupled CPU‑GPU architecture will be the most efficient way to meet the projected compute surge.

Bottom line

Lisa Su’s bold numbers are a reminder that the AI boom is not just a headline; it is a structural shift that will reshape hardware demand for the next decade. AMD is positioning itself to capture a slice of that demand through a mix of silicon, software, and a growing engineering footprint in Asia. Whether the industry can actually scale to 10 yottaFLOPS remains to be seen, but the trajectory set by AMD and its peers suggests that the compute crunch will be a defining challenge—and opportunity—for the entire tech ecosystem.

Read the full transcript of AMD’s AI Developer Day on the official AMD newsroom.