Definition

Intel

What is Intel?

Intel Corp. is the world's largest manufacturer of central processing units and semiconductors. The company is best known for CPUs based on its x86 architecture, which was created in the 1980s and has been continuously modified, revised and modernized.

Intel also offers graphics processing units (GPUs), networking accelerators, programmable processors, AI accelerators, and communications and security products.

Intel’s CEO is Pat Gelsinger, one of the most significant engineers in the company’s history. He co-designed the 80386 processor, led the development of the 80486 processor, served in a variety of roles including the first chief technical officer and created the Intel Developer Forum tradeshow.

He was pushed out of the company in 2009 following a power struggle. He went to EMC and later VMware, where he rose to the role of CEO. He returned to Intel as CEO in 2021.

History of Intel 

Intel was founded in 1968 by a group of semiconductor engineers from Fairchild Semiconductor, led by Gordon Moore and Robert Noyce. The company's name comes from shortening the term integrated electronics.

Although the company created its first CPU in 1971, Intel's primary business was static random access memory (SRAM) and dynamic random access memory (DRAM) chips. IBM used the Intel 8-bit chip 8088 processor when it created the PC in the 1980s, and then moved to Intel's 16-bit 8086.

IBM was hesitant to be dependent on a single supplier for its CPU, so it required Intel to license the 8086 architecture to other chipmakers, including AMD, Cyrix, Via Technologies and STMicroelectronics. Intel's disputes over licensing its technology with AMD eventually led to the two companies going to court, with the judge ruling in favor of AMD. Intel was forced to license its technology, which remains in place today.

Intel continued its x86 naming convention with the 80286, 80386 and 80486 processors. In 1993, it changed its brand name to Pentium.

In 1998, Intel expanded its focus from desktops to server processors with the introduction of its Xeon line. Before this, server processors were manufactured mainly by Unix vendors such as IBM and Sun Microsystems.

In the mid-1990s, Hewlett-Packard partnered with Intel to make a new family of processors derived from HP's PA-RISC called Itanium. While the Itanium was 64-bit and had instruction sets that were favorable to mission-critical computing, its lack of compatibility with x86 code severely limited its appeal. The Itanium was capable of x86 emulation, but performance was unacceptably slow. In addition, HP was the only significant OEM customer for the processor and developing applications on it was difficult. Intel released the last iteration of the Itanium processor in 2017, canceled development in 2019, and officially ended sales of the product line in 2023.

In 2004, Intel decided to license AMD’s 64-bit instruction set rather than develop its own instructions, which could have taken two or three years. The primary benefit of 64-bit processing is memory capacity; the most memory a 32-bit processor could access is just 4 GB, while a 64-bit processor can access up to 16 exabytes of memory.

In 2005, Apple CEO Steve Jobs announced the company would transition its Macintosh line of computers from Motorola's PowerPC to Intel's x86, ending a 20-year relationship between Apple and Motorola. In 2020, Apple announced that it would make its own CPUs, ending its 15-year relationship with Intel.

In 2006, Intel addressed its muddied and convoluted product line with a brand called Core, replacing the Pentium naming scheme. Under the Core brand name are four numbers indicating level of performance --- 3, 5, 7, and 9, with 3 being the lowest performance and 9 the highest.

Later in the year, Intel introduced the Tick-Tock manufacturing cadence. One year would be designated a “tick” year marked by a manufacturing shrink (from 90nmto 65nm, for example) while the next year would be a “tock” year where it introduced an entirely new microarchitecture. Intel would follow this roadmap for nearly 20 years before design and manufacturing challenges proved difficult to keep the yearly cadence.

In 2007, Apple approached Intel about building a mobile processor for a smartphone it had in development. Intel passed on the opportunity, which would become the iPhone. After the release and subsequent success of the iPhone, Intel quickly produced a mobile processor called Atom, an alternative to Apple's Arm mobile processor. Intel's submission would prove too late, however, and the company would wind down Atom development and its pursuit of the smartphone market in 2016.

In 2021, Intel introduced the integrated device manufacturing (IDM) 2.0 initiative, a radical departure from the way it manufactured previous processors. Up to this point, Intel made all of its own chips and almost never made chips for competitors. Under IDM 2.0, it would partner with other leading semiconductor manufacturing firms to outsource some of its chip manufacturing, most notably Taiwan Semiconductor Manufacturing Corporation.

The second part of the IDM strategy is known as Intel foundry service, where Intel manufactures chips for fabless semiconductor makers. Due to the expense of maintaining and updating fabrication facilities, most semiconductor firms -- including Nvidia -- have their chips manufactured by third-party foundry companies.

In 2021, Intel joined the chip fabrication market to capitalize on the growing need for foundry capacity. Its initial customers include AWS and Qualcomm.

Intel products and services

Intel manufactures and offers a multitude of products and services in the computing space.

Core processor

Intel's top product is its CPU line for desktop and laptop computers. Primary products are sold under the Core brand name, with the newest generation of Core i9 processors having up to 24 cores.

Low-end processors are still sold under the Celeron and Pentium brand names as of 2024, with Celerons and Pentiums coming in dual- and quad-core editions. They are primarily targeted at very low end, entry-level systems.

Supporting PC chipset

In 2007, Intel introduced its vPro brand for remote IT management and support. It comes as standard issue on every Intel motherboard for a server, laptop and desktop computer.

The vPro business technology enables IT departments to monitor, update and troubleshoot multiple PCs without having to physically go to the hardware in question. IT professionals can remotely troubleshoot a laptop in another office.

Xeon processors

In 1998, Intel decided to jump into the server business with the Xeon processor. The Xeon processor is built on the same basic CPU design as the Core desktop and laptop processor, but it offers extra technology for reliability, scalability and availability. It's designed for much less fault-tolerant use cases, such as servers.

Intel has six different families of Xeon processors:

  • Xeon D. For small systems when space and power are at a premium.
  • Xeon E. For business-ready performance in entry server products.
  • Xeon W. For creative professionals using high-end workstations and running VFX, 3D rendering and 3D CAD.
  • Xeon Scalable. Design for high-end servers in data centers. Xeon Scalable is broken down into four separate product lines:   
    • Xeon Scalable Bronze -- entry-level performance.
    • Xeon Scalable Silver -- mid-range performance.
    • Xeon Scalable Gold – high-performance with additional features.
    • Xeon Scalable Platinum -- premium performance for the most mission critical workloads.
  • Xeon Phi. Meant to compete with GPUs in the high-performance computing and parallel processing space, Intel has faltered with this product line and reduced emphasis on it.
  • Xeon Max. The successor to the Xeon Phi, this is a high-performance compute offering featuring advanced capabilities for specialized applications.

GPUs

Intel has had GPU technology for a long time, but it was only for integrated graphics. Intel had low-end GPUs as part of its Core processors, which are suitable for running Windows and productivity applications. The company didn't compete with superior GPU cards from Nvidia and AMD in the gaming or enterprise markets.

However, this changed in 2018 when Intel announced development of its Xe GPU architecture and its intent to compete with Nvidia and AMD in the PC graphics card market. Released in 2020, Xe is found in consumer products sold under the Arc brand name and in enterprise GPUs under the codename Ponte Vecchio. However, Intel has struggled against the deeply entrenched Nvidia and AMD Radeon GPUs and rarely discusses its consumer GPU efforts anymore.

High-performance computing and AI

Intel has a variety of processors for AI and high-performance computing, including the Xeon Scalable CPU, Alteryx field-programmable gate arrays and the Ponte Vecchio GPU. Intel also had a dedicated AI processor called Habana, which it acquired after purchasing Habana Labs in 2019 for $2 billion. Habana performs both training and inference in AI. This is important as training and inference are two separate processes with different compute requirements.

In 2024, Intel discontinued the Habana line in a strategic shift to streamline its AI processor line. The last processor from Habana -- called Gaudi 3 -- is focused primarily on AI training and competes with the GPUs from Nvidia and AMD.

After that, Gaudi technology will merge with Xe GPU technology to form a new line of AI processors code-named Falcon Shores.

Optane persistent memory

Optane is a unique, Intel-only technology that serves as a cache for SSDs. SSDs are faster than HDDs, but still considerably slower than standard DRAM. Intel's Optane has almost the same speed as DRAM, but it also the persistence of an SSD. Optane sits between flash storage and DRAM and acts as a cache for SSDs. It's a server–only technology and caters to the enterprise market.

Optane never gained any traction outside of a few OEMs, and Intel discontinued the product line in mid-2023.

oneAPI

OneAPI is a unifying API for all of Intel's semiconductors. OneAPI decides the best processor for the application and compiles it for that processor. This frees the developer from having to worry about low-level programming tasks. As of this writing, oneAPI is still under development and Intel promises to release it as open source.

FPGA

In 2015, Intel shelled out $16.7 billion to acquire Altera, a maker of field programmable gate array (FPGA) processors. FPGAs are known for specializing in parallel processing and being reprogrammable or reconfigurable for different use cases and scenarios, something a CPU cannot do.

In 2024, Intel spun off its FPGA unit under the brand name of Altera with its own CEO. The purpose of the spinoff was to let the FPGA group operate with autonomy separate from the CPU business. Intel felt that it could not operate its FPGA business unit like it did its CPU unit and it needed to be independent.

This was last updated in August 2024

Continue Reading About Intel

Networking
Security
CIO
HRSoftware
Customer Experience
Close