Qualcomm's "three key players" have left, and Intel has welcomed a GPU leader
Intel recently announced the appointment of Eric Demers, a former senior GPU architect at Qualcomm, as its chief GPU architect, marking a key move in the advancement of its GPU strategy.
Against the backdrop of the continuous upgrading of the global artificial intelligence (AI) and high-performance computing race, graphics processing units (GPUs) have become a core area of competition among global semiconductor giants as the core computing power support. For a long time, NVIDIA has formed a monopolistic pattern in the high-end GPU and AI computing power field by virtue of its mature chip architecture and complete industrial ecosystem.
As a long-established leading enterprise in the semiconductor industry, Intel is clearly unwilling to lag behind in this race for computing power. In recent years, it has continuously intensified its strategic layout in the GPU field, striving to break through the existing market competition barriers and seize the core discourse power in the computing power race. Recently, Intel announced the appointment of Eric Demers, a former senior GPU architect at Qualcomm, as its chief GPU architect, marking a key move in the advancement of its GPU strategy.
Intel's core objective in laying out its strategy in the GPU field is to build a complete computing power ecosystem of "CPU + GPU + AI accelerator", and thereby form a competitive tripartite pattern with NVIDIA and AMD in the industry. To achieve this strategic goal, Intel has taken the lead in launching Xe architecture independent GPU products, comprehensively covering multiple application scenarios such as consumer level, professional level and data center level.
Beyond its independent GPU layout, Intel has subsequently released the Gaudi series of AI accelerators, with a focus on the large model training and inference market. Among them, Gaudi 3, as an upgraded product in the series, has achieved significant improvements in manufacturing processes and core performance. Theoretically, it has strong competitiveness in terms of inference performance and cost-effectiveness for large language models (LLMs), and has received technical and market support from mainstream server suppliers such as Supermicro and Dell, as well as IBM.
However, Intel's GPU strategic planning process was not without challenges. Due to various subjective and objective factors, the market acceptance of Gaudi 3 did not reach the expected level after its launch. The Falcon Shores project, which was originally planned to integrate the graphics processing capabilities of Xe with the AI performance advantages of Gaudi, was also announced to be terminated. All these further delayed the overall progress of its GPU business.
To address the shortcomings of its current business, Intel introduced the Gaudi 3 hybrid AI rack solution at the 2025 OCP Global Summit and reached a deep strategic partnership with NVIDIA. NVIDIA subscribed to Intel's common stocks worth 5 billion US dollars. Both parties plan to jointly develop customized CPUs based on the x86 architecture and deeply integrate them with GPUs through NVIDIA's NVLink high-speed interconnection technology, aiming to create a new generation of heterogeneous computing platform for data centers and AIPC.
Intel's CEO Chen Liwu stated that the Intel GPU project will focus on AI and data centers. Demers will serve as the company's senior vice president and will report directly to Kevork Kechichian, the executive vice president and general manager of the Data Center Group. Of course, "the company has also made a lot of efforts to invite Demers to join."
Demers also later posted on social media, stating that over the past few months, he had multiple conversations and meetings with Chen Liwu. He was deeply impressed by Chen Liwu's confidence and optimistic attitude, and was also excited about working at the new Intel and participating in this ongoing transformation.
As a top talent with decades of experience in the semiconductor industry, Demers worked at Qualcomm for 14 years. Before leaving, he held the position of Senior Vice President of Engineering and was the core leader in the research and development of Qualcomm's Adreno GPU series. He led the full lifecycle evolution of this architecture and promoted the wide application of Adreno GPUs in various terminal scenarios such as mobile phones, PCs, automobiles, and XR, providing solid technical support for Qualcomm to consolidate its advantages in graphics performance of mobile chips.
Before joining Qualcomm, Demers had accumulated nearly a decade of industry experience at AMD/ATI. He served as the chief architect for ATI Radeon R300 and R600 architectures, as well as the CTO of AMD's graphics department. He also participated in the design of the graphics module for Nintendo's GameCube. He is one of the few top talents in the industry who can independently build a GPU architecture, possessing both technical depth and industry vision.
It is worth noting that during the period when Demers switched from Qualcomm to Intel, Qualcomm also lost two key CPU architects - Gerard Williams III, who led the development of the self-developed Oryon CPU architecture, and John Bruno. Currently, neither of them has disclosed their future work plans.
As a veteran professional in the semiconductor chip field, Gerard Williams III has a career spanning three major industry leaders: Arm, Apple, and Qualcomm. During his tenure, he led the development of classic architectures such as Arm Cortex-A8 and Cortex-A15. While working at Apple, he served as the chief architect for the A12X to A17 series chips, holding over 60 related technology patents and directly driving the technological breakthroughs and industrialization of Apple's self-developed chips.
In 2021, Qualcomm acquired NUVIA for 1.4 billion US dollars. Gerard Williams III also officially joined Qualcomm and took on the position of Senior Engineering Vice President. During his tenure at Qualcomm, he led his team to optimize and upgrade the Phoenix architecture from the NUVIA era into the Oryon architecture. This architecture, as Qualcomm's first fully self-developed high-performance CPU architecture, has become the core technology foundation of the Snapdragon series of high-end chips.
As a co-founder of NUVIA, John Bruno joined Qualcomm along with Gerard Williams III and held the position of Engineering Vice President. His main responsibilities included the microarchitecture design, performance optimization, and power management of the Oryon CPU. He was a key driver in bringing this architecture from prototype development to large-scale mass production.
It is widely believed in the industry that the departure of these two architects is more due to adjustments in their personal career发展规划. It is a temporary phenomenon in the flow of industry talents. With its mature technical system and complete industrial layout, Qualcomm will continue to consolidate its core advantages in mobile terminals and chips for multiple scenarios.
The financial performance report of Qualcomm for the first quarter of the fiscal year 2026 shows that the company's revenue for this quarter reached 12.25 billion US dollars, an increase of 5% year-on-year; the adjusted net profit was 3.781 billion US dollars, a decrease of 1% compared to the previous year; the adjusted earnings per share was 3.50 US dollars, slightly higher than the 3.41 US dollars in the same period last year. The overall performance exceeded market expectations.
The quarterly revenue of the core chip business QCT was 10.613 billion US dollars, an increase of 5% year-on-year. Among this, the revenue from mobile phones was 7.824 billion US dollars, a 3% increase year-on-year. The revenue from automobiles was 1.101 billion US dollars, a 15% increase year-on-year, exceeding 1 billion US dollars for the second consecutive quarter. The revenue from IoT was 1.688 billion US dollars, a 9% increase year-on-year, maintaining a stable trend.
Looking ahead, Qualcomm expects its second-quarter revenue to be between 10.2 billion and 11 billion US dollars; and its adjusted earnings per share to be between 2.45 and 2.65 US dollars. Meanwhile, the analysts surveyed by LSEG predict that Qualcomm's second-quarter revenue will be 11.11 billion US dollars and its earnings per share will be 2.89 US dollars.
BEIJING 
ENGLISH 
USA 
DEUTSCH 
FRENCH 
JAPAN 
SPANISH 
AUSTRALIA
