OpenAI’s GPT-6 Astra Could Boost CPU Demand and Benefit Intel and AMD

Key Takeaways

• OpenAI has introduced GPT-6 Astra, describing it as its most capable and aligned model, with leading performance in computer use, software engineering, cybersecurity and scientific research.
• Astra can navigate software, complete multi-step workflows and interact with browsers, spreadsheets and desktop applications, potentially increasing the computing resources required to run AI agents.
• Secure containers, enterprise orchestration systems and parallel software testing could generate additional CPU workloads, creating a potential demand tailwind for Intel and AMD.
• The CPU opportunity remains an industry projection rather than confirmed demand directly attributable to Astra, while GPUs will continue to handle most model training and inference.

OpenAI’s latest artificial intelligence model could broaden the semiconductor industry’s AI opportunity beyond graphics processors, potentially creating additional demand for CPUs supplied by Intel and Advanced Micro Devices.

The company introduced GPT-6 Astra last week, calling it its most intelligent and aligned model to date. OpenAI said Astra delivers state-of-the-art performance across computer use, browsing, software engineering, cybersecurity, scientific research and professional work.

The model is initially being deployed to a limited number of organizations, with access expanding to ChatGPT Plus, Pro, Business and Enterprise users, as well as the OpenAI API, Microsoft Azure and Amazon Bedrock. OpenAI President Greg Brockman reportedly described the launch as the beginning of an artificial general intelligence era.

While Astra’s advanced reasoning continues to rely heavily on GPU-powered cloud infrastructure, analysts believe its ability to autonomously operate software and coordinate multiple agents could also increase demand for general-purpose processors.

That would potentially benefit Intel and AMD, two companies that have received less attention than Nvidia during the first phase of the generative AI investment boom.

GPT-6 Astra Moves AI From Answering Questions to Completing Work

Unlike a conventional chatbot that primarily generates text in response to prompts, GPT-6 Astra can execute multi-step tasks across different applications.

OpenAI says the model can fill out forms, update customer records, conduct online research, organize calendars, create websites, analyze scientific data and produce finished documents, spreadsheets and presentations. It can also install software, run tests and troubleshoot problems displayed on a computer screen.

Astra achieved a score of 72.6% on the OSWorld 2.0 computer-use benchmark, compared with 65.7% for GPT-5.6 Sol. OpenAI also reported that Astra completed simulated computer-use tasks in approximately 47% less time than its predecessor.

These capabilities allow an AI agent to interact directly with graphical user interfaces. Instead of requiring a dedicated application programming interface for every piece of software, the model can navigate applications in a manner closer to a human user.

The shift is important for hardware demand because performing work generates a much larger number of computing operations than simply producing an answer. Each task may involve opening applications, retrieving information, executing code, validating results and launching additional agents.

Why GPT-6 Astra Could Increase CPU Demand

The potential CPU benefit is based on three parts of the model’s agentic workflow.

Secure Virtual Environments

OpenAI has classified Astra at the Critical cybersecurity capability threshold under its Preparedness Framework. During testing without production safeguards, the model achieved a 100% score on ExploitBench and discovered two previously unknown software vulnerabilities.

Because of those capabilities, companies may deploy Astra inside virtual machines, sandboxes or secure containers that isolate the model from sensitive production systems.

Creating, managing and terminating large numbers of isolated environments requires CPU cores, memory and operating-system resources. If multiple agents are running simultaneously, the infrastructure requirements could grow rapidly.

OpenAI has introduced monitoring and authorization controls for Astra, while advanced cybersecurity tasks remain restricted. Enterprise administrators must also enable the model before it can be used in their organizations.

Enterprise Data and Orchestration

Companies using Astra with proprietary data may need to operate supporting frameworks that connect the model to databases, internal applications and document repositories.

Although the model’s main inference workload can remain in the cloud, the surrounding orchestration layer may run inside an enterprise data center, private cloud or local workstation. That layer handles authentication, retrieval, application access, data processing and communication between agents.

These are varied, general-purpose workloads that are normally handled by CPUs rather than GPUs.

AMD has argued that agentic AI will increase demand for processors because agents must perform database queries, tool calls, validation checks and other operational tasks after a model generates its instructions. The company describes GPUs as the AI system’s reasoning engine and CPUs as the infrastructure responsible for executing and coordinating the resulting work.

Parallel Agents and Software Testing

Astra can divide a large objective into smaller tasks and assign them to multiple agents. A coding project, for example, could involve separate processes for analyzing source code, testing alternative solutions, compiling programs, checking security vulnerabilities and validating a user interface.

Running these activities in parallel can create substantial CPU demand. Unit tests, software compilation, browser sessions and local development tools typically depend heavily on general-purpose processors.

Early reports from Astra users suggest the model is willing to launch numerous agent processes when handling complex assignments. If that behavior becomes common in enterprise deployments, companies may need servers and workstations with more CPU cores, additional memory and higher input-output capacity.

How Intel Could Benefit

Intel could benefit through both its Xeon server processors and its client computing business.

Enterprise Astra deployments may require Xeon-powered servers to operate secure containers, coordinate multiple agents and connect AI models to corporate applications. Intel has increasingly positioned the CPU as the execution and control layer within heterogeneous AI systems.

The opportunity could also extend to commercial PCs. If AI agents run continuously on employee devices, businesses may need to replace older computers with systems offering faster processors, larger memory capacity and dedicated AI acceleration.

This could support demand for Intel’s Core Ultra and workstation processors, although the scale of the benefit will depend on whether Astra workloads are primarily executed locally, in private data centers or through public cloud services.

Intel has not publicly provided a revenue forecast tied specifically to GPT-6 Astra.

How AMD Could Benefit

AMD is similarly positioned to capture demand across servers, workstations and AI-enabled personal computers.

The company says its EPYC processors are designed to support the large number of concurrent tasks created by multi-agent systems. Its sixth-generation EPYC portfolio offers as many as 256 cores and 512 threads per socket, providing the density required to run numerous agent processes simultaneously.

AMD has also introduced rack-scale AI systems combining EPYC CPUs with Instinct accelerators. The company said AI is increasing demand across data centers, PCs and edge devices, potentially expanding its total addressable market to approximately $2 trillion by 2030.

OpenAI and AMD are already working together to optimize GPT-class workloads across AMD’s software and hardware platforms. OpenAI expects to begin using AMD’s Helios infrastructure in the fourth quarter of 2026, with deployments accelerating in 2027.

Astra could strengthen this opportunity if enterprises begin deploying agentic workloads at scale, but AMD has not disclosed any orders directly connected to the model’s launch.

CPUs Will Complement Rather Than Replace GPUs

The potential increase in CPU demand does not reduce the importance of GPUs.

Large language models still require accelerators for training, inference and intensive mathematical calculations. CPUs perform a different role by coordinating agents, managing operating systems, running software tools and processing the actions produced by the model.

Astra could therefore expand the overall AI hardware market instead of shifting spending entirely from one processor category to another.

Nvidia may continue to capture the largest share of spending on model training and inference, while Intel and AMD benefit from the supporting execution infrastructure. Cloud providers, memory manufacturers and networking companies could also gain if AI agents generate more continuous and distributed computing activity.

Trade global stocks through CFDs on Markets.com and take positions on price movements without owning the underlying shares. Eligible new clients can also explore up to $5,000 in bonuses, subject to the applicable promotion terms.

The CPU Opportunity Is Not Yet Guaranteed

The investment case remains dependent on how customers ultimately deploy Astra.

If most agent activity occurs inside centralized cloud platforms, the demand benefit may be concentrated in data-center CPUs rather than consumer PCs. Software improvements could also make agent workflows more efficient, reducing the amount of hardware required for each task.

Companies may limit the number of concurrent agents because of security, cost and governance concerns. Astra’s cybersecurity capabilities could encourage heavily isolated deployments, but strict controls might also slow enterprise adoption.

There is also no direct evidence yet that Astra has produced a measurable increase in Intel or AMD processor sales. The CPU demand argument is based on the model’s technical capabilities and the expected infrastructure requirements of large-scale agentic AI.

Nevertheless, Astra reinforces a broader change in the AI market. As models move from generating content to independently operating software, more computing work takes place outside the model itself.

If businesses begin running thousands of autonomous agents across applications, databases and secure environments, CPUs will become increasingly important to the AI infrastructure stack. That development could give Intel and AMD a larger role in the next stage of the AI investment cycle.