Analyst(s): Dr. Bob Sutor
Publication Date: September 6, 2024
CMU-connected startup Efficient’s partnership with GlobalFoundries (GF) is notable for such a young company and speaks well of its quick move to packaging and manufacturing. Energy efficiency is vital for many reasons, but not all are related to sustainability in the computer industry.
What is Covered in this Article:
- Efficient, its energy-efficient CPU, and the GF partnership
- The reasons why we care about energy efficiency and sustainability with processors
- GlobalFoundries and its national and New York State deep tech ecosystem
The News: On August 29, Pittsburgh startup Efficient announced a strategic partnership with GlobalFoundries (GF) to manufacture a new high-performance computer processor that Efficient claims is up to 166x more energy-efficient than industry-standard embedded CPUs.
GlobalFoundries Advances Energy Efficiency with Efficient Partnership
Analyst Take: The race to produce new processors is well underway, especially in the AI space, where power-hungry accelerators for training and inference focus on CPUs, GPUs, and specialized chipsets. Yes, the vendors focus on saving energy, but often so it can be used elsewhere. The small CPUs operating in low-power environments are at the other end of the spectrum. This is Efficient’s sweet spot, and moving from stealth in March to a partnership with GF in August demonstrates significant momentum.
What Is Efficient and What Is Its Chip?
Efficient was founded in January 2022. Its CEO is Professor Brandon Lucia of Carnegie Mellon University (CMU) in Pittsburgh, Pennsylvania, USA. Many of its technical staff have graduate degrees from CMU and Stanford.
In March 2024, Efficient emerged from stealth mode and announced its low-power processor architecture. It also stated that it received $16M in seed funding from California venture capital firm Eclipse. Eclipse is also an investor in AI accelerator chip manufacturer Cerebras Systems. At the time, Efficient stated its chip was up to 100 times more energy efficient than leading in-market general-purpose CPUs. The new announcement has increased that to 166 times more energy efficient than “industry-standard embedded CPUs” based on a 5-by-5 convolution benchmark. This is a machine learning test, so Efficient is not simply looking at common CPU applications but also future ones that will come with the increasing use of AI.
It’s not just a single processor that will deliver commercial value: the chipset packaging will use MRAM and the Adaptive Body Biasing (ABB) energy-leak prevention capabilities of GF’s 22FDX® platform. It is in this sense that GF is truly a partner of Efficient and not just its fab of choice.
Whatever the multiple and type of CPU, these impressive results will improve existing products and make viable new ones for which the power consumption was previously too high.
Why Do We Really Care About Chip Energy Efficiency?
In the vernacular, “sustainable energy” and “energy-efficient” are sometimes incorrectly used as close synonyms, though the former is much broader and certainly extends well beyond chips, computers, and data centers. Sustainable energy is a source of energy that cannot be depleted under typical situations and is therefore available for future generations. The use rate is relevant to depletion, so we care about making our systems energy-efficient. This is one reason why we are concerned with the amount of energy semiconductors use and the heat and other by-products they produce. These may have further energy costs to remediate.
A second reason is the total energy budget. If we cap the total energy available to a system or data center, we must examine the energy consumption of components and processes. We then optimize the entire configuration so we can accomplish what we require. If we replace a CPU or a GPU with one or more that use less energy and have at least the same processing power, we may be able to use the extra power available for other infrastructure. The result is more computation using the same or less energy.
The optimization is not limited to processing units. Hard disk drives (HDDs) were first commercialized in 1957 by IBM and are still in use. However, solid-state disk drives (SDDs) are smaller and use less energy for comparable storage but are more expensive. Here, we have a tradeoff of size and energy reduction versus cost. The complete energy optimization of a system has many variables and data points.
A third reason for semiconductor energy efficiency is also related to the power budget but on the small end of the spectrum. For example, if we run on batteries, we may only have a few watts or fractions to power everything. Using less power extends the battery lifetime. IoT, edge processing, and personal smart devices such as watches and medical devices are good example use cases.
There is much information on the web about sustainable energy and energy-efficient systems. For a comprehensive textbook, see Kutscher, C.F.; Milford, J.B.; Kreith, F. (2019). Principles of Sustainable Energy Systems. Mechanical and Aerospace Engineering Series (Third ed.). CRC Press. ISBN 978-0-429-93916-7. Also, see the UN 2030 Agenda for Sustainable Development.
GlobalFoundries in the Upstate New York Deep Tech Ecosystem
GF was founded in 2009 and was formed with the spinoff of AMD’s semiconductor manufacturing business. In 2014, IBM sold its Microelectronics Division, including its New York and Vermont fabs, to GF.
In February 2024, the US government awarded GF $1.5B in potential investment under the CHIPS and Science Act. New York State awarded GF $620M. GF will use the funds to modernize existing facilities, expand the manufacturing fab to support the US auto industry, and build a new state-of-the-art fab for AI chips and industry-specific semiconductors.
In June, Australian startup Diraq chose GF to fabricate its cryogenic quantum processing units.
Though incorporated in the Cayman Islands, GF is headquartered in Malta, New York, about 30 miles north of the state capital, Albany, and 180 miles north of New York City. Thus, while it is in New York State, it is in upstate New York. It is one of the eastern anchors of what I call the “New York Deep Tech Thruway.”
The Albany NanoTech Complex and Albany AIM Photonics integrated photonic circuit center are also at the eastern end of this innovation thruway. GF, too, is a leader in silicon photonics. Photonics technology is essential to computer and communications networking and several quantum computing modalities, including trapped ions and neutral atoms.
Moving westward, the Quantum Information Science Research Center of the US Air Force Research Laboratory is in Rome, and Micron Technology plans to spend up to $100B over the next two decades to build a semiconductor factory complex north of Syracuse.
AIM Photonics has a test, assembly, and packaging facility in Rochester. The city is home to the Rochester Institute of Technology and the University of Rochester, notably its Institute of Optics.
Finally, we get to Lake Erie. In January, the University at Buffalo announced its Center for Advanced Semiconductor Technologies.
As the US recovers from its critical semiconductor supply issues during the Covid pandemic, the CHIPS and Science Act and industry, educational, and workforce initiatives from New York State are visibly changing the future landscape for classical, edge, AI, photonic-based, and quantum computation. The GF partnership with Efficient is evidence of this and is an excellent collaboration.
What to Watch:
- Can GF and Efficiency get samples to customers in summer 2025?
- Will GF increase its focus on energy efficiency and sustainability?
- Will GF and New York State companies get additional grants from the US CHIPS and Science Act or state programs?
- Will Efficient be acquired by another processor manufacturer looking for a market advantage in the low-power space?
For additional details, see the “Efficient and GlobalFoundries Partner to Enable a New Category of Ultra Energy-Efficient, High-Performance Processors” press release.
Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.
Other insights from The Futurum Group:
Talking Marvell, GlobalFoundries, Google, CrowdStrike, Intel, OpenAI
GlobalFoundries Reports Second Quarter 2024 Financial Results
Author Information
Dr. Bob Sutor is an expert in quantum technologies with 40+ years of experience. He is the accomplished author of the quantum computing book Dancing with Qubits, Second Edition. Bob is dedicated to evolving quantum to help solve society's critical computational problems.
