It’s an exciting time for Intel. It’s been a long time since CIOs, CTOs, and IT leaders have considered Intel a top strategic IT partner. Even though its silicon was embedded throughout their environment, they didn’t really need to think about it, as silicon wasn’t a major strategic decision.

AI changes all this. Tech leaders are rethinking their strategic partnerships because silicon matters again. Whether your infrastructure is in the cloud, on-premises, at the edge, or any combination thereof, the accelerators and specific features in your silicon can make an enormous difference to what you can do with AI and how quickly you can roll it out.

The Intel Vision 2024 conference on April 8 and 9 was filled with its latest strategic vision and announcements. Intel dedicated much of the event to presenting its vision as a company (a systems and foundry company) and its vision for “AI Everywhere.” It supported these efforts by detailing its competitive advantages and demonstrating that it’s growing an ecosystem committed to partnering with Intel. Here are some key takeaways from the Intel Vision event for tech leaders:

  • Intel is evolving from a silicon company to a systems company. Intel’s vision is to bring to market a comprehensive set of software, services, and solutions delivered as a system. The focus is on systems where silicon plays a critical role. Today, Intel’s software and services portfolio is limited. Silicon-relevant software includes Intel’s Edge Platform, which runs across all its hardware offerings and allows enterprises to secure, deploy, and manage highly distributed edge implementations. For services, Intel Trust Authority provides a Zero Trust attestation SaaS offering for confidential computing environments.
  • Intel is betting that tech leaders will want to embed AI everywhere … From one end of the computing spectrum (PCs) to the other (superclusters), Intel’s strategy is to embed features and capabilities that support AI. Its slate of current and upcoming chips support: 1) AI PCs (Core); 2) AI processing and decision-making at the edge; 3) smaller, narrower large language models (LLMs) on enterprise server CPUs; and 4) very large LLMs that leverage GPUs in clusters and superclusters. Intel Foundry provides the additional capability of providing customers with custom AI chips that fit their specific use cases.
  • … and that they will need different levels of AI at different price points. Intel predicts that the market is going to require different levels of AI capabilities — and one that they have invested in. It’s a safe bet; not every company needs to train and deploy very large LLMs. For many organizations, it makes more sense to deploy smaller, narrower LLMs and retrieval-augmented generation (RAG) for specific tasks — hence, why smaller models and RAG, tuned to your organization needs, can be much better; you can train it with minimal data, and it’s more accurate and cost-effective. For these AI workloads, Intel’s Xeon processor offers several advantages, such as power efficiency at a lower price compared to NVIDIA’s high-end Blackwell processor.
  • Intel is working hard to build a large ecosystem on an open foundation. Throughout the event, Intel invited a nonstop parade of enterprise clients, cloud and infrastructure partners, SIs, AI software startups, etc., to the stage. Intel strongly believes that its commitment to open models, open data, and open-source tools will fuel an ecosystem of companies and organizations looking for an alternative to NVIDIA’s more expensive, proprietary approach to silicon. There were plenty of small, innovative startups such as NAVER and LANDING AI alongside heavy weights like Accenture, Dell, Infosys, and Red Hat. Intel also continues to promote its own OpenVINO software.
  • Intel hopes to speed up the detection and response aspects of security. Intel is looking to decrease the time between threat detection and threat prevention. This shows up in several efforts: 1) at endpoints, with the use of the existing threat detection technology (TDT) within some Intel CPUs; 2) in explorations into the new neural processing unit (NPU) in AI PCs to do local, rapid analysis of fileless malware; and 3) embedding into security providers’ cloud architecture as these companies look to add copilot functions to their SaaS-based solutions.

Despite much of the focus being on vision, partners, and strategic goals, there were also notable product announcements, including:

  • Xeon 6. The venerable server chip family targeted at data centers, cloud, and the edge with Efficient-core (E-cores, code-named Sierra Forest) and Power-core (P-cores, code-named Granite Rapids) variants. Processors with E-cores focus on efficiency and serve as replacements for older systems to reduce energy consumption and promote sustainability. Processors with P-cores can run LLMs such as Llama 2 with the right configuration.
  • Gaudi 3. This is Intel’s new accelerator that is performant, power-, and cost-efficient at scale. It is set to compete with NVIDIA and AMD, and as such, Intel has pulled in several partners like Bosch, Dell, IBM, NAVER, and many others. Look for Intel to build its reputation and credibility in the general-purpose GPU (GPGPU)/AI accelerator market with Gaudi 3.
  • Intel Tiber portfolio of business solutions. This includes the portfolio of software solutions and support services that aim to achieve outcomes higher up the tech stack. For example, American Airlines reports saving 23% on data lake costs with Intel Tiber App-Level Optimization. There are four main categories in the Tiber portfolio today: 1) AI software (Intel Developer Cloud, AI Studio); 2) trust and security (Trust Services, Transparent Supply Chain); 3) cloud (App-Level Optimization); and 4) Edge (Intel Edge Platform).

For more perspective from us (Alvin Nguyen, Paddy Harrington, Michele Pelino, Stephanie Balaouras, and Glenn O’Donnell) or any other Forrester analyst, book an inquiry or guidance session at inquiry@forrester.com.