Oxford connects different atom species across a quantum network
Two network nodes combine strontium links with calcium storage. The four-atom demonstration makes the division of labor visible.
Two elements, one shared state
An Oxford team linked strontium and calcium atoms across two quantum-network nodes, creating shared states involving up to four atoms. The Physical Review Letters paper, published October 13, reports storage of remotely memory for about ten seconds. The underlying preprint was first submitted June 17, 2025; October marks the journal milestone.[2][3]
Different jobs can favor different hardware
The experiment assigns different roles to the two elements: strontium participates in the optical network link, while calcium supplies the longer-lived storage described by the researchers. That is an intuitive engineering idea with a demanding quantum implementation. A system can become more capable by connecting specialized parts, but the connection must preserve the information that makes those parts useful. An interface that destroys the relevant quantum relationship would defeat the purpose.[1]
The importance of waiting
Distributed computing rarely proceeds with every required resource ready at exactly the same instant. A quantum network has an additional constraint: a useful state must survive while other work catches up. Generating a link and keeping it usable are therefore separate capabilities. A longer memory is valuable only in relation to the timing and error budget of the operation that will eventually consume it. The smallest successful network experiments can make those relationships visible.
A network milestone with a clear scale
This remains a controlled small-node demonstration, not a worldwide quantum internet. Its value is more specific: it brings multiple species, remote entanglement and storage into the same experiment. QubitWire’s interpretation is that the next compelling comparisons should show how often usable states are produced, how long they remain useful and what operation follows. A single maximum lifetime is interesting; a dependable schedule for creating and consuming network resources would be more informative still.
For newcomers, this is a good illustration of why the future computer may contain several kinds of quantum hardware rather than one universal-looking chip.
Compare the two clocks
The paper reports that storage lasts more than 100 times as long as creating the entangled state. That comparison explains why ten seconds can matter in this apparatus. It does not mean the network completed 100 useful operations: other operations, errors and scheduling would have to be counted to make that claim.[2]
Memory and computation answer different questions
A separate Oxford study published in February 2025 shows the complementary task. Across two modules about two metres apart, the researchers teleported a with 86% fidelity and ran a distributed Grover search with a 71% success rate. Those older numbers describe an earlier computation experiment, not the performance of every state in the October memory paper.[4]
Together, the studies suggest a practical way to read the next network headline: ask what was connected, how long the shared state remained useful and which calculation actually consumed it. Improving any one of those stages matters. A larger usable network requires them to work together under the same operating conditions.
Research in context · checked 20 Oct 2026
Why a long-lived link can still be a slow link
A worked explanation of the distinction behind the result.
Suppose a hypothetical network needs two separate connections before it can perform its next step. The first becomes available at time 2 and the second at time 8, measured in arbitrary units. The first connection must remain useful for six units while waiting. If it does not, the network has to start some work again. A component can be excellent at producing an occasional link and still struggle in this combined schedule.
Now give the first connection a memory whose probability of survival after a waiting time t is exp(-t/T). This is a deliberately simple model. When T equals six units, surviving the wait has probability exp(-1), about 36.8%. Increasing T changes the schedule’s prospects, but so does producing the second link earlier. The model shows why memory lifetime, link generation time and coordination belong in the same discussion. It is not a forecast for the Oxford apparatus.
A fixed lifetime does not mean the state is perfect until a stopwatch expires and useless immediately afterward. Reliability can degrade gradually and differently for different observables. A complete experiment needs a success criterion, not just a duration. The criterion might concern the fidelity of the final shared state or whether a later operation meets its requirement.
This suggests a more informative network dashboard: a timeline showing when each resource appears, how its quality changes while waiting and when it is consumed. Such a display could help readers understand why a modest memory improvement may make a complicated protocol easier, while a much larger headline improvement might contribute little if another step dominates.
The central question is therefore not which component owns the biggest number. It is whether the components can work together before their shared resource becomes inadequate.
Sources and evidence
- Quantum network entangles several atoms of different elements ↗Source published 2026-10-15. university announcement.
- Multipartite Mixed-Species Entanglement over a Quantum Network ↗Source published 2026-10-13. peer-reviewed research.
- Distributed quantum information storage across a network of trapped-ion nodes ↗Preprint first submitted June 17, 2025; the October journal publication is a later milestone.
Research in context · checked 20 Oct 2026
Does entanglement let a network send messages instantly?
A worked explanation of the distinction behind the result.
Entanglement does not provide a controllable faster-than-light message channel. A shared quantum state produces correlations, but turning a network protocol into useful communication still requires ordinary information exchange and defined local operations. Drawing a glowing line between two atoms can conceal that distinction.
A network diagram becomes clearer when quantum and classical links have different labels. The quantum connection supplies a resource; the classical connection carries coordination information. This also helps distinguish remote entanglement from simply synchronizing two classical devices. The remarkable part is the particular joint state that can be prepared and preserved, not the removal of ordinary communication constraints.
The shape of the waiting-time distribution can matter as much as its mean. A protocol that occasionally waits much longer may lose stored resources in those rare intervals. Whether those events are tolerable depends on how the network retries, schedules work and defines a successful output.
Sources and evidence
- Quantum network entangles several atoms of different elements ↗Source published 2026-10-15. university announcement.
- Multipartite Mixed-Species Entanglement over a Quantum Network ↗Source published 2026-10-13. peer-reviewed research.
- Distributed quantum information storage across a network of trapped-ion nodes ↗Preprint first submitted June 17, 2025; the October journal publication is a later milestone.