New quantum milestone: IBM reaches 433 qubits with the new Osprey processor, more capacity than the number of atoms in the universe

From left to right, Darío Gil, vice president and head of IBM Research;

Jay Gambetta, of the IBM Thomas J Watsonand Research Center, with the Osprey chip in hand;

and Jerry Chow, director of the Experimental Quantum Computing group, at the company's quantum center in New York.IBM

“A quantum computer with 273 qubits has more memory than atoms in the observable universe”, according to Alberto Casas, researcher at the CSIC and author of

(Ediciones B, 2022).

The physicist predicted in his book an imminent achievement of this milestone and it has already been achieved.

The multinational IBM has presented this Wednesday its new processor Osprey (Osprey), with which it has reached 433 cubits, more than triple the capacity of that achieved only a year ago with the Eagle (127 cubits) and a tenth of the they expect in 2025. It is a necessary step to develop effective quantum computers, capable of performing large-scale practical calculations in a robust way, without errors.

Classical physics does not explain our world, although until now it has served us well, especially in the macroscopic realm.

Quantum science, which studies nature and interactions at the atomic and subatomic scale, begins to provide possible explanations and maintains two paths that intersect: the theoretical one, which aspires to answer fundamental questions, such as what is the dark matter that makes up the universe, and the practice, which wants to apply quantum laws to processes such as computation to exponentially multiply the ability to study and understand life and everything that surrounds it.

This last path has surpassed a new goal.

More information

Alberto Casas' fundamental guide to understanding quantum physics

A qubit is the minimum unit of quantum computing and, unlike the conventional bit —the one used in common technology—, it does not have only two values ​​(0 and 1), but can have these or any superposition

from them.

If a current supercomputer can do millions of byte operations (IBM's Summit is capable of processing more than 200 billion calculations per second), a quantum one can perform trillions.

Hence the importance of increasing the capacity of qubits in the models under development and minimizing the errors that affect this technology.

This is one of the achievements of IBM, which maintains its roadmap to achieve an effective quantum computer.

The milestone announced this Wednesday, which has been achieved with innovations in programming (

) and

(physical or material elements of computers and computer systems), is the basis for the future launch of the Condor, which will be the first processor universal quantum of more than 1,000 qubits.

This will be followed by the Kookaburra (Cookaburra), scheduled for 2025 and which will have a capacity of more than 4,000.

According to IBM CEO Arvind Krishna, this latest computer "will be able to perform calculations that would require a traditional computer almost the size of the Earth."

In this race for the "quantum advantage", a term that refers to developments based on this physics that surpass classical computers, there are also companies like Google or Rigetti.

To achieve this, IBM is not only focusing on quantum properties, but also includes conventional processors based on their efficiency in computing tasks.

The goal is to develop a circuitry fabric to distribute complex problems across multiple processors.

Quantum System TwoIBM

In this way, to triple the capacity of the Eagle (the processor that surpassed the 100 qubit barrier), the new Osprey is included in a new IBM Quantum System Two, a modular and flexible system to incorporate multiple chips (integrated circuits) connected by a control system.

This will be accessible online in the first quarter of 2023 and is decisive for developing a modular architecture with quantum communication —which increases computational capacity— and with a hybrid cloud system that allows the integration of quantum and classical workflows.

"We have

finalized what the system will look like from a design perspective and we are working to build it, put it in place and demonstrate it next year,” says Jay Gambetta, the physicist who

leads the IBM Thomas J Watson Research Center team for the development of the quantum computer.

It will be the basis for the Crossbill, the first single processor made from multiple chips, and the Flamingo, which will include quantum communication links.

Both developments are scheduled for 2024. A year later, with the combination of the two technologies, the Kookaburra is expected to be presented, a quantum system with three processors and 4,158 qubits that aims to open the door to overcome the 100,000 qubit barrier.

the noise problem

But to take advantage of the computational advantage of the qubit, a difficult problem must be overcome.

The quantum superpositions of states that allow the trillions of combinations and thus the almost infinite computing capacity are very sensitive to the environment.

Any little environmental circumstance (temperature, electromagnetic noise or vibration) degrades the overlays and generates errors.

Jian-Wei Pan, China's top computer expert, says that "building a practically usable and fault-tolerant quantum computer is one of the great challenges for human beings."

“The most formidable obstacle is the presence of noise and imperfections.

We need to use quantum error correction and fault-tolerant operations to overcome noise and scale the system,” he says.

IBM Quantum Osprey Processor.IBM

One of the formulas to alleviate this disadvantage is to create environments where interactions with the environment are minimized, such as cooling systems to a temperature close to absolute zero (-273 ºC).

But errors can also be addressed once they occur, by applying systems, such as the Qiskit Runtime Primitives that incorporates the Osprey presented this Wednesday.

This development is an environment of classical and quantum systems that increases the speed and quality of computing.

In addition, Dynamic Circuits is added, an alternative model of circuit construction for quantum error correction.

IBM plans to introduce new fault suppression and mitigation systems to help developers of the core operating system program (kernel) manage noise and fix bugs.

Compared to material alternatives, Qiskit Runtime Primitives allows solutions against noise to be programmed, which makes it easier, according to the company, for "users to incorporate quantum computing into their workflows and accelerate the development of quantum applications ”.

Oliver Dial, head of quantum

architecture at IBM, believes that they are not very far from minimizing the noise: “In the next two years, we will be able to do something that nobody has done before.

It is a challenge.

If we can provide an estimate that is free of observable noise, we will be in the range where we can start solving interesting problems with our customers.”

In this way, according to IBM, the quantum race that is taking hold will not require having a computer with this technology, but rather a developer will be able to incorporate certain quantum calculation functionalities and operations into their programs that will be executed in the cloud and that will be integrated with fluency in applications that will combine new and existing technology.

A 4,000-qubit computer will be able to perform calculations that would require a traditional computer almost the size of the Earth.

Arvind Krishna, CEO of IBM

Alejandro González Tudela, scientific researcher at the Institute of Fundamental Physics of the CSIC, believes that the possibilities of quantum computing are extraordinary and that many of them are yet to be discovered.

One of the fields of application will be, in his opinion, “questions of many bodies, with many elements that interact with each other and that are difficult to solve in classical computers”.

Two examples would be the simulation of molecules or the development of new materials, fields with an exponential dimension unattainable for traditional systems.

Artificial intelligence (AI) and, in particular, neural networks, which try to emulate the human brain, will also benefit from this technology to classify, analyze and extract patterns and knowledge from images, words or concepts of any type of language in any area.

In this sense, IBM has a system capable of interpreting the language of chemistry to predict the most likely result of a given reaction.

This convergence between bits, neural networks and qubits is the technological foundation of a new era of discoveries and a revolutionary future of innovations for science as well as for companies and institutions.

You can write to us at