The Leap Second Is Going. Your Computer Clock Will Notice.

Delegates at the General Conference on Weights and Measures have voted to end leap seconds from May 20, 2027, removing a source of disruption for computer systems worldwide. ^[2][3]

The decision, passed by a non-unanimous majority on October 15, 2026, accelerates a change agreed in 2022 that would have ended the adjustments by 2035. Instead of keeping the international time standard close to time based on Earth's rotation, the new arrangement will allow a gap of up to one hour before a correction is needed. ^[2][3]

That does not mean an extra hour will appear on clocks in 2027. It means the world will stop making occasional one-second adjustments and accept a much wider separation between the time kept by atomic clocks and the time measured by the turning Earth. ^[2][3]

For networks and data centres, the attraction is continuity. For astronomy, the trade-off is a looser connection between civil time and the planet whose rotation originally gave the day its meaning. ^[2][3]

Georgette Macdonald said the decision would preserve the continuity and relevance of Coordinated Universal Time, or UTC, for centuries. Noël Dimarcq described it as a decisive step towards an international reference time scale that could remain continuous for millennia. ^[2]

One second can be expensive

UTC is the atomic-clock-based standard used to coordinate time internationally. A separate scale, UT1, follows Earth's rotation. Leap seconds have kept those two measures close by occasionally adjusting UTC by one second. ^[2][3]

That adjustment is tiny on a wristwatch. Across computer systems trying to agree on when something happened, it creates an awkward exception to the normal progression of time. Nature documented computing disruptions and clocks falling out of alignment on the day of an adjustment. ^[2]

The consequences have reached well beyond specialist timekeeping. The 2012 leap second was associated with crashes at Reddit and LinkedIn and delays to Qantas flights, according to Michelle Bastian, a senior lecturer at the University of Edinburgh. ^[3]

Meta, Google, Microsoft and Amazon supported ending the practice, according to The Verge. That support predates the final vote and should not be read as a statement of how each company will implement the new rule. ^[1]

The engineering appeal is straightforward: remove an exceptional clock adjustment, and systems no longer have to accommodate that particular exception. Ending leap seconds cannot make every clock agree or prevent every outage. It does remove a known source of trouble from the shared standard. ^[2]

The decision also brings the deadline forward substantially. Organisations working towards the earlier 2035 target now have a May 2027 start date for the new arrangement. The policy question has been settled much sooner than the timetable agreed four years ago suggested. ^[1][3]

Keeping a common clock

Some companies had already found their own ways around the leap second. Google, for example, spread an extra second across 20 hours rather than applying the whole adjustment at once, a practice called smearing, according to Bastian. ^[3]

That is an elegant solution to a local engineering problem. It also exposes a problem for an international standard: if organisations handle the same adjustment differently, agreeing to use UTC does not necessarily mean their clocks behave identically through the adjustment.

Bastian argues that preserving UTC as a common standard helped drive the decision. The concern was that organisations developing their own workarounds could weaken its role as a shared reference. On that reading, the vote is also an attempt to keep users together by changing the standard they share. ^[3]

A standard has little practical authority if its users find it too troublesome to follow. Removing leap seconds gives technology companies something they wanted, but the benefit of a common clock extends beyond those companies. The value lies in different organisations being able to rely on the same reference, rather than each solving timekeeping alone.

The majority vote nevertheless stops short of showing universal agreement. It passed despite opposition, according to Nature. The choice involves competing priorities, even if the computer failures make the practical case for change easy to grasp. ^[2]

A wider gap from the Earth

The cost is that UTC will no longer stay as closely aligned with Earth's rotation. Bastian argues that astronomy loses that close link between civil time and the rotating planet. The one-hour allowance makes the scale of the policy change clear: the permitted difference will be far larger than the one-second adjustments being retired. ^[2][3]

An atomic clock and a measure of Earth's rotation answer different questions. The old arrangement kept their answers close. The new one gives priority to continuity in the international clock, while allowing the astronomical measure to move farther away. ^[2][3]

The detailed method and timing of any eventual correction have not been established publicly in the accounts of the decision. Nor have the voting breakdown, operators' implementation plans or guidance for astronomical users been specified. ^[2][3]

For time services, operating systems, networks and data centres, the next task is to translate the decision into procedures. Google's earlier workaround shows why that work matters: changing the international rule and changing how systems handle it are separate jobs. ^[3]

The date those preparations must address is May 20, 2027, when leap-second adjustments are due to stop. ^[2][3]