Advanced Micro Devices concluded its 2025 fiscal year with robust performance, reporting optimistic revenue growth and significant margin improvement. The company has outlined an ambitious roadmap, planning to launch the MI400 series GPUs in the second half of 2026, followed by the MI500 series in fiscal year 2027. As AI developers and enterprises begin expanding dedicated AI inference compute capacity, Advanced Micro Devices is positioned to enter a new growth trajectory. Bolstered by a solid financial outlook for 2026 and a substantial stock price correction following the Q4 2025 earnings release, analysts have upgraded the stock rating to Strong Buy.
In the fourth quarter of 2025, Advanced Micro Devices delivered impressive sales figures driven by its MI350 series GPUs and EPYC CPUs, with net revenue surging 37.58% year-over-year. Fiscal year 2025 saw particularly strong growth in EPYC instance deployments, with cloud service provider deployments increasing over 50% and enterprise deployments more than doubling. A key highlight of Q4 2025 was Advanced Micro Devices' next-generation Instinct series AI accelerator GPU portfolio: the MI455X and Helios designed for AI supercomputing clusters; the MI430X for high-performance computing and sovereign AI deployments; and the MI440X targeting enterprise customers for both AI training and inference.
To support this, Advanced Micro Devices will partner with Hewlett Packard Enterprise to introduce a rack-scale architecture solution utilizing Broadcom's Tomahawk 6 networking chips for elastic scaling networks. The server solutions will incorporate HPE's customized hardware and software solutions based on the Ultra Accelerator Link open Ethernet standard, alongside direct liquid cooling infrastructure. A single server rack can connect 72 MI455X GPUs, representing an enterprise-ready solution expected to drive substantial growth.
The company anticipates the MI450 series GPUs will launch in the second half of 2026. Management also announced the next-generation MI500 series GPUs are scheduled for 2027, promising a 1000-fold AI compute improvement over the MI300X series, utilizing a 2-nanometer process and featuring HBM4E high-bandwidth memory. For Q1 2026, management expects China sales of MI308 series GPUs to reach $100 million, while actively pursuing export licenses for the MI325 series, which recently received approval from the U.S. Department of Commerce. However, new export regulations requiring priority supply for U.S. domestic needs introduce some uncertainty regarding the scale of Advanced Micro Devices' China sales expansion.
Another notable development in Q4 2025 was Advanced Micro Devices' agreement with Riot Platforms for 25 megawatts of data center capacity, expandable to 200 megawatts over the contract term. Although the company didn't disclose the specific reason for the compute purchase, analysts speculate it might involve deploying rack-scale solutions for private cloud development services or internal research and development.
Additional opportunities may arise from Advanced Micro Devices' relationship with OpenAI. A Reuters report from February 2, 2026, indicated OpenAI's dissatisfaction with the performance of the latest AI chips from its primary supplier, noting the AI firm had been seeking alternatives since 2025. Shortly after the report, OpenAI CEO Sam Altman stated on social media platform X, "We are very happy to work with our primary supplier, who makes the world's leading AI chips, and hope to remain a major customer long-term." This tension is particularly evident with OpenAI's Codex code assistant product, which demands extremely responsive compute power. Subsequently, the primary supplier announced plans to invest $20 billion in OpenAI, a significant 80% reduction from the initial $100 billion proposal in September 2025, sending a clear signal.
Beyond a previously agreed 6-gigawatt compute collaboration established in October 2025, the challenges facing the dominant supplier could present a growth opportunity for Advanced Micro Devices. Concurrently, Intel announced on February 3, 2026, its return to the GPU market, appointing a chief GPU architect to lead the project. Notably, Intel had already partnered with the dominant supplier to develop a custom CPU series for its NVLink solution. This development could impact Advanced Micro Devices' enterprise and cloud AI growth narrative in two ways: Intel's GPUs might target the enterprise inference market, and the combination of the dominant supplier's GPUs with Intel CPUs could be deeply optimized for high-end AI compute. However, with Intel's 14A process volume production not scheduled until next year, it is too early to determine the final direction of this architecture.
While analysts believe the partnership between the dominant supplier and OpenAI is unlikely to see major disruption, competition in the AI accelerator chip market is expected to intensify significantly throughout 2026. In early January 2026, OpenAI also partnered with startup Cerebras to access 750 megawatts of low-latency AI compute. Cerebras utilizes a large-chip design approach that could potentially address inference compute bottlenecks.
In the client and gaming segment, Advanced Micro Devices' semi-custom chips are in the early stages of production ramp-up for Microsoft's next-generation Xbox console, scheduled for 2027. Despite the potential market appeal of the new console, management has guided for a double-digit revenue decline in the semi-custom business for 2026.
Advanced Micro Devices' operational efficiency improved markedly in Q4 2025, with margins expanding across the Data Center, Client, and Gaming segments, although the Embedded segment margin declined by 164 basis points. For Q1 2026, analysts project net revenue of $9.9 billion and adjusted earnings per share of $1.36. Benefiting from the volume ramp of MI350 series products, adjusted gross margin is expected to improve further in Q1 2026; however, margin may face temporary pressure during the production ramp of the MI400 series in the second half of 2026.
The company's financial foundation remains solid, with a balance sheet showing $10.5 billion in cash and cash equivalents against $3.22 billion in debt, resulting in a net cash position of $7.33 billion. This provides significant financial flexibility for potential mergers and acquisitions and internal growth investments. Inventory saw a slight increase in Q3 and Q4 2025, primarily due to the MI350 series production ramp and the reversal of inventory write-downs for the MI308 series.
In an optimistic scenario, as more AI developers focus on inference performance optimization, Advanced Micro Devices could capture additional data center GPU market share. The urgent need for high-speed compute in enterprise services like code development, combined with OpenAI's push for supplier diversification, could allow Advanced Micro Devices to gain further traction in the AI inference market. A pessimistic scenario considers the growing importance of custom chips for AI inference, exemplified by Google's TPU chips and its collaboration with Anthropic, suggesting rising competitive barriers within the industry.
Advanced Micro Devices currently trades at a price-to-sales ratio of 9.56, representing a relatively high valuation premium compared to its one-year trading range. Despite short-term premium valuation, analysts believe the stock still possesses upside potential as the company capitalizes on data center growth, particularly the emergence of AI inference demand. Following the Q4 2025 earnings report, Advanced Micro Devices' stock price experienced a significant decline, dropping 17% the next day. Although market sentiment turned negative, analysts attribute this drop largely to external factors rather than a deterioration in the company's fundamental operational performance. From a fundamental perspective, Advanced Micro Devices now presents a more attractive investment opportunity.
Comments