ASML's time-stamp service concept cut extreme chip-tool outages 20%

ASML's chip-making machines are so expensive that unplanned downtime at a customer's fab is measured in lost semiconductor output. The company's service network keeps over 10,000 different components and modules in warehouses worldwide.

At the start of 2017 ASML began rethinking its after-sales service supply for the rare but disruptive long-downtime events that customers feared most. The new concept adds a 'time stamp' — an estimate of how long a part or tool takes to travel from the service network to a specific broken machine — and changes planning accordingly.

After successful pilots, the concept was rolled out to the worldwide installed base in 2020. The result: a 20 percent decrease in extreme long-downtime events, worth an estimated €1.5 billion a year to the semiconductor industry.

The work was recognized as a runner-up in the 2021 INFORMS Wagner Prize competition for operations-research practice.

Guarantee time to part, not part count

Time-to-part, not stock availability, is what a fab actually cares about.

Focusing on rare extreme events targets the biggest cost, not the average.

A worldwide rollout proved the concept at industrial scale.

An estimated €1.5B yearly industry benefit made the case to customers.

For machines where a downtime day costs millions, time-to-part for rare failures beats average availability; redesigning the service promise around it cut the costliest outages by a fifth.

ASML rolled the new service-supply concept out worldwide in 2020; extreme long-downtime events fell 20 percent, worth an estimated €1.5 billion a year for the semiconductor industry, and the work was a 2021 Wagner Prize runner-up.