At IonQ, we are actively developing new vacuum solutions for future systems that have the potential to reduce computational energy costs for systems powerful enough to deliver commercial quantum advantages, all while meeting enterprise-grade customer requirements. This blog will discuss our progress to date in vacuum research, the path ahead, and why solving energy demands is so critical for the quantum industry.
The Computational Energy Crisis
The advent of generative artificial intelligence has brought unprecedented computer-generated experiences to the world and transformed many people's understanding with regard to what computers are capable of doing. Behind these experiences is an unfortunate truth: classical techniques are increasingly energy-hungry. A proliferation of larger and larger models, which require more and more energy to hone and improve the large language models driving this revolution. In the United States alone, the increase in power demand due to data center demand is expected to rise from 200 TWh as of 2022 to 260 TWh in 2026, equivalent to 6% of all power use across the country1 . One could argue that this issue will be solved by developing more energy efficient GPUs or by transitioning energy hungry data centers into more climate friendly energy sources, like nuclear power. However, at IonQ, we believe that these incremental approaches to the exponentially accelerating problem the industry faces will fall short.
IonQ is hard at work inventing a new approach to computing. One that taps into the quantum properties of naturally occurring atoms to perform new types of calculations to solve problems more efficiently and to open up a new class of problems previously deemed impossible to solve. As we work tirelessly towards this goal, we feel it is important to learn from the past and work to sidestep the same pitfalls the classical industry now faces. Currently, the world's largest classical computer corporations are grappling with the clash of their public energy commitments and the world’s thirst for increasingly powerful computational workflows. IonQ is already thinking of how to avoid a similar fate for quantum computing.
The Potential Quantum Computing Energy Impact
To say broadly that quantum computing is more energy efficient than classical computing would be misleading. This is because there are many different approaches to building quantum computers currently being explored, and the energy requirements for different approaches can vary drastically. The one commonality across all quantum computing modalities is the need to highly isolate qubits from the surrounding environment. Quantum information is famously fragile and highly susceptible to interference from the outside world. The most common technique for superconducting and photonic architectures to increase isolation and induce quantum properties is to create both extremely cold and extremely high vacuum environments for the qubits to operate in. While cooling may be done relatively efficiently, the energy required to accurately control the qubits in this extreme environment is intense and scales directly with the size of the quantum computer. In 2023, IBM stated that their flagship quantum computer required 35 Watts per qubit2 . If IBM is able to deliver on the 10,000 qubit system on their roadmap at the current energy requirements, the system would require 3.5 Mega-Watts to operate. 3.5 Mega-Watts could power roughly 3,000 average American homes3 . It's easy to see how such an approach could lead to a recreation of the energy crisis the classical computer industry is facing.
At IonQ, the primary mechanism to achieve isolation for our atomic qubits is through vacuum. We share this in common with other trapped ion approaches as well as neutral atom architectures. To get a sense of the vacuum levels being applied to quantum computers, see the chart below:
However, even within this corner of the quantum neighborhood, there are important differences in how the vacuum is achieved and maintained. Generally, the approaches being utilized in quantum computing today fall into the following categories:
Unlike superconducting systems, which control qubits inefficiently through electronic means, IonQ uses photons to connect the classical control systems to the qubits within the vacuum chamber. This connection is highly energy efficient and results in a considerably lower energy requirement to run quantum calculations. Furthermore, this control system scales very efficiently; a single AOD control system can scale to many qubits with the same energy profile.
This leaves our vacuum chamber as a core area of research and development to both improve qubit performance through isolation and to lower our energy requirements by applying an efficient vacuum. Over the years, IonQ has experimented with various vacuum approaches in our commercial systems. Our first commercial system, IonQ Harmony, used a room-temperature vacuum chamber. Subsequent systems, IonQ Aria and IonQ Forte applied closed-loop cryogenics to improve vacuum and, consequently, qubit performance. Our key vacuum research thrust is to design a system with the best of both worlds: extreme-high vacuum and room temperature operation.
Building Enterprise-Grade Quantum Computers
With these seemingly extreme environments being pursued within quantum computing, it’s easy to see how requirements for deploying these systems may vary drastically from classical computer requirements. However, our pursuit of room-temperature vacuum chamber research is evidence of IonQ’s unique approach to building quantum computers. While we are highly focused on developing larger and higher-performance quantum computers, perhaps most importantly, we are doing that with an eye to our current and future customers’ enterprise needs.
We are diligent in not compromising customer needs in the pursuit of scale and performance, and we call this Enterprise-Grade.
XHV Research at IonQ
Our current systems are attaining XHV through the use of cryostats. Recently, we have invested in new technologies that would enable a miniaturized, room temperature, XHV. If successful, we believe the downstream benefits to our customers are numerous.
To date, we have completed two key steps in building this solution. First, we have developed a manufacturing chamber. This chamber is capable of reaching ultra-high vacuum (UHV) and is big enough for our team to assemble vacuum chambers within.
The assembly chamber is located in our Seattle manufacturing facility, which was opened in February 2024. The assembly chamber is equipped with manipulators needed to hermetically seal miniaturized ion trap packages within UHV conditions – vacuum within vacuum. In doing so, several pumping stages may be skipped, and only pumping mechanisms directed at XHV pressures are needed. This allows the form factor of the ion trap package to be reduced considerably. Integrated heaters allow for an effective bake-out of the ion trap package components – elevating the components to high temperatures to drive out trapped gases that often limit obtainable vacuum levels. The components of the ion trap package are loaded in and later removed from a load lock chamber; a special technique borrowed from the semiconductor industry that allows the main assembly chamber to permanently remain at UHV pressures.
IonQ UHV Assembly Chamber
Next, we designed a miniature ion trap package to assemble in the manufacturing chamber. The ion trap packages are engineered to be able to achieve XHV without moving parts. This is made possible through careful choices of vacuum pumps and materials. Gettering technology is used to provide high pumping rates, using passive components that dissipate no power. An ion pump captures those gases not pumped by the getters. Only very clean materials are used, and each component undergoes rigorous qualification before use – similar to studies done at NASA, CERN, and LIGO. In addition, only the highest quality of hermetic joints are used to seal the packages, and each one must be thoroughly qualified using precision leak detection technology.
While achieving XHV is an endeavor in its own right, simultaneously operating an ion trap presents an outstanding challenge. Integrated into the ion trap packages are optical windows for manipulating ions using lasers-cooling, detection, and qubit operations. A specialized electrical feedthrough allows for the seamless operation of the ion trap’s many electrodes without disturbing vacuum quality. The end result is the core of an ion trap quantum computer, and you can hold it in your hand.
The Path Forward
Up next, we’ll be trapping ions and establishing baseline operating conditions for our compact packaging. We’ll continue to tweak our designs and processes in the push towards XHV. And that’s just the first step–we’ll continue working towards the full vision of this technology and have multiple exciting milestones ahead of us in order to begin using this technology for quantum computing applications.
1 https://www.forbes.com/sites/arielcohen/2024/05/23/ai-is-pushing-the-world-towards-an-energy-crisis/↫
2 https://travislscholten.substack.com/p/podcast-summary-oliver-dial-ibm-quantum↫
3 https://www.eia.gov/energyexplained/use-of-energy/electricity-use-in-homes.php↫