Leap seconds end in May 2027: what continuous UTC changes

Leap seconds now have an end date. The General Conference on Weights and Measures has decided that continuous Coordinated Universal Time will take effect on May 20, 2027, removing the periodic adjustments used to keep civil time closely aligned with Earth's rotation. The change gives digital infrastructure a more predictable common clock while leaving astronomers with a separate measurement of how the planet turns.

The 2026 resolution raises the permitted difference between UTC and rotation-based UT1 to 3,600 seconds. That is a long-term limit on their divergence. It does not instruct phones, computers or railway clocks to jump forward by an hour next May.

Two ways to measure time

An atomic clock and a rotating planet answer different questions. One provides a highly regular interval; the other tells us Earth's orientation. Keeping their answers close has required occasional intervention in the international time scale used as the basis of civil time.

BIPM's explanation of UTC describes a reference assembled from roughly 450 atomic clocks across about 80 laboratories. International Atomic Time, or TAI, is continuous. UTC runs at the same rate but differs by a whole number of seconds. Laboratories maintain local realizations, called UTC(k), which are compared through BIPM's published results.

That arrangement is more subtle than one master clock broadcasting a definitive reading from a single room. International timekeeping combines measurements, comparisons and local delivery. Its value is the common reference that different institutions can reproduce and relate to one another.

Why continuity matters to computers

The resolution identifies a particular concern: a negative leap second has never been applied, and critical infrastructure is poorly prepared for one. It also points to inconsistencies caused when service providers handle leap adjustments in different ways. A workaround that smooths a change locally can leave two otherwise accurate services using slightly different time during the adjustment.

Consider two systems recording opposite ends of one transaction. If they use different interpretations of the same clock correction, comparing their timestamps becomes harder precisely when an engineer needs to reconstruct the sequence. Continuous UTC removes a source of that disagreement; it cannot make every machine's clock accurate or eliminate delays between machines.

The distinction is important because a stable standard and perfect synchronization are different achievements. Devices still need reliable time sources, and applications still need to account for the conditions under which timestamps were produced. Removing exceptional steps simplifies the shared reference rather than abolishing the engineering around it.

Earth's rotation remains part of the picture

Users who need Earth's orientation will still need UT1. The decision calls for continued monitoring, with the difference available through established services. Astronomical work does not become less interested in the planet's actual position because ordinary civil clocks stop making tiny corrections to follow it.

BIPM's background on time metrology, published before this meeting, distinguishes continuous UTC from another project: preparing a future definition of the second using optical frequency standards. One concerns how the time scale proceeds; the other concerns how its fundamental unit may eventually be defined. They should not be conflated into a claim that seconds become longer or shorter in May.

That separation between a measurement and the standard built around it is central to the science of timekeeping. Atomic clocks can supply a regular interval while observations continue to describe the less regular planet. Civil time no longer has to force both answers into such close agreement.

For everyday readers, the result should be continuity. For organizations comparing events across machines, the payoff is a common clock with one fewer exceptional behavior to reconcile.