Why GPU-only systems aren’t enough for AI agents

AI Inference Using CPU Why GPU-Only Systems Are No Longer Sufficient for Autonomous AI Agents

2026-04-13 From Manuel Christa | Translated by AI 2 min Reading Time

Related Vendor

tmts2026-ticec-banner-en-2000px (https://www.tmts.tw/en)

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

AI applications are increasingly working independently. However, pure graphics processors are reaching their limits. Intel and Sambanova are now developing an architecture that combines different chip types.

Infrastructure for AI agents: Intel and Sambanova share computing load across three processor types(Image: Gemini / AI-generated) — Infrastructure for AI agents: Intel and Sambanova share computing load across three processor types
(Image: Gemini / AI-generated)

When AI no longer just generates texts, but writes code and queries databases independently as an agent, previous hardware structures come to a standstill. AI systems based on graphics processing units (GPUs) work inefficiently with these complex tasks. To resolve this bottleneck, Intel and Sambanova have presented a blueprint for future server infrastructures. From the second half of 2026, they want to offer data centers and cloud providers hardware that specifically divides up the computing work.

The concept is based on a heterogeneous architecture. Instead of burdening a single chip type with all phases of AI calculation, the partners assign the appropriate processor to each step.

Special Chips for Each Calculation Phase

The design provides for a precise division of tasks. GPUs take over the starting signal: in the prefill phase, they process long input texts into usable data records. Then the specialized RDU chips from Sambanova take over. They generate the responses of the language models and ensure a high data throughput during decoding.

Intel contributes the Xeon 6 server CPU as a control center. This coordinates the tasks of the AI agents and distributes the workload. At the same time, the Intel chip acts as an executing element that compiles and checks the code. Rodrigo Liang, head of Sambanova, summarizes: "Agent AI is going into production—and the success pattern we see is GPUs to start the job, Intel Xeon 6 to run it, and Sambanova RDUs to finish it quickly."

Software Compatibility As A Basis

Compatibility is a key advantage of the architecture. As Intel supplies the host processors, companies can continue to use their x86 software environment without any adjustments. No single type of chip is optimal for every phase of a workflow. Intel and Sambanova's blueprint stands out because it combines high performance with fewer chips and full compatibility with the existing software ecosystem.

Systems in comparison: According to the manufacturers, the architecture from Intel and Sambanova for large AI models requires significantly fewer chips and less power than competing solutions and is easier to integrate into existing data centers.(Image: Sambanova) — Systems in comparison: According to the manufacturers, the architecture from Intel and Sambanova for large AI models requires significantly fewer chips and less power than competing solutions and is easier to integrate into existing data centers.
(Image: Sambanova)

Sambanova's own measurements support this approach. According to the company's data, the Xeon 6 compiles code more than 50 percent faster than server CPUs based on the Arm architecture. The Intel chip also works up to 70 percent faster with vector databases than comparable competitor products from the x86 camp. The cooperation between the two companies thus marks a clear step from pure feasibility studies to the broad commercial use of AI agents in data centers. (mc)

Subscribe to the newsletter now

Don't Miss out on Our Best Content

Business E-mail

Please enter a valid mailadress.

By clicking on „Subscribe to Newsletter“ I agree to the processing and use of my data according to the consent form (please expand for details) and accept the Terms of Use. For more information, please see our Privacy Policy. The consent declaration relates, among other things, to the sending of editorial newsletters by email and to data matching for marketing purposes with selected advertising partners (e.g., LinkedIn, Google, Meta)

Date: 08.12.2025

Naturally, we always handle your personal data responsibly. Any personal data we receive from you is processed in accordance with applicable data protection legislation. For detailed information please see our privacy policy.

Consent to the use of data for promotional purposes

I hereby consent to Vogel Communications Group GmbH & Co. KG, Max-Planck-Str. 7-9, 97082 Würzburg including any affiliated companies according to §§ 15 et seq. AktG (hereafter: Vogel Communications Group) using my e-mail address to send editorial newsletters. A list of all affiliated companies can be found here

Newsletter content may include all products and services of any companies mentioned above, including for example specialist journals and books, events and fairs as well as event-related products and services, print and digital media offers and services such as additional (editorial) newsletters, raffles, lead campaigns, market research both online and offline, specialist webportals and e-learning offers. In case my personal telephone number has also been collected, it may be used for offers of aforementioned products, for services of the companies mentioned above, and market research purposes.

Additionally, my consent also includes the processing of my email address and telephone number for data matching for marketing purposes with select advertising partners such as LinkedIn, Google, and Meta. For this, Vogel Communications Group may transmit said data in hashed form to the advertising partners who then use said data to determine whether I am also a member of the mentioned advertising partner portals. Vogel Communications Group uses this feature for the purposes of re-targeting (up-selling, cross-selling, and customer loyalty), generating so-called look-alike audiences for acquisition of new customers, and as basis for exclusion for on-going advertising campaigns. Further information can be found in section “data matching for marketing purposes”.

In case I access protected data on Internet portals of Vogel Communications Group including any affiliated companies according to §§ 15 et seq. AktG, I need to provide further data in order to register for the access to such content. In return for this free access to editorial content, my data may be used in accordance with this consent for the purposes stated here. This does not apply to data matching for marketing purposes.

Right of revocation

I understand that I can revoke my consent at will. My revocation does not change the lawfulness of data processing that was conducted based on my consent leading up to my revocation. One option to declare my revocation is to use the contact form found at https://contact.vogel.de. In case I no longer wish to receive certain newsletters, I have subscribed to, I can also click on the unsubscribe link included at the end of a newsletter. Further information regarding my right of revocation and the implementation of it as well as the consequences of my revocation can be found in the data protection declaration, section editorial newsletter.