Telstra blames missing software update and undocumented design change for massive outage

Vicki Brady told senators maintenance staff were unaware of mass outage risk as critical software failure ‘rippled slowly across the network’

Follow our Australia news live blog for latest updates

Get our breaking news email , free app or daily news podcast

Telstra has blamed the lack of a software update on a key time-keeping system for an outage that caused nationwide chaos last week, with its maintenance teams also unaware of a design change that affected how it would reset.

Telstra’s chief executive, Vicki Brady, told a Senate inquiry into the mobile outage on Friday that 45% of all calls and data sessions last Wednesday were affected when the network went down shortly before 4.30am AEST.

Telstra said it did not lack redundancy in its network – but that such redundancy did not prevent the outage.

In the submission and hearing, executives confirmed reporting that one of Telstra’s three network time protocol (NTP) servers, designed to ensure the systems had the correct time, had reset to 2006 .

Brady identified the server as a Microchip model called SSU 2000. It was manufactured in 2011, and costs $30,000 to replace. Telstra has three in Sydney, Melbourne and Perth. It was still under support from Scientific Devices but the inquiry heard the issue was not hardware – but that the telco had not applied a software update.

During maintenance to replace faulty backup power, the company shut down and restarted the server in Melbourne. Due to an “underlying software configuration” within the server, it restarted with the wrong 2006 date.

“Over the next few hours, the incorrect date rippled slowly across the network, causing authentication certificates in other servers to become invalid,” Telstra said.

“Customers were intermittently unable to authenticate onto the network (‘no service’), which affected their ability to place voice calls and use data across Telstra’s mobile network.”

The executives revealed that in addition to the server losing its ability to communicate with servers in other locations, in both 2022 and January this year, the company was alerted by the manufacturer that it needed to update the software. Had the update been applied, the outage may have been avoided, they said.

Telstra had also made an intentional design change to the equipment to fix an earlier fault – but this had not been properly documented for the maintenance workers to be aware of.

The company’s executive for end-to-end resilience, Gerard Tracey, also said that had Telstra invested in a new piece of hardware operating in the same design it was intended to the outage may not have occurred.

The submission also said when the Melbourne NTP server disconnected for maintenance, the other two worked as redundancy and backup as expected.

“The failure mode here was not inherently related to hardware, levels of redundancy, or the architecture of our network,” Telstra said.

But when the Melbourne server supplied an incorrect date once switched back on, “downstream systems used that date in security, authentication, session and policy-control processes.”

“The issue was therefore not simply the loss of one NTP server or redundancy in the design of the configuration of the three NTP servers, but the propagation and acceptance of erroneous date information by interconnected systems that rely on timing as a trust and ordering reference.”

Telstra said the cause of the outage was unacceptable, and suggested the controls were not good enough. Brady told the inquiry Telstra was taking full accountability for the outage.

“Last week, Telstra let Australians down, we let our customers down, we let the community down, and we fell short of what people rightly expect from us,” she said.

Telstra said its investigation will probe why the design change was not documented, and the software update not completed.

In the course of the outage, Telstra said 58,835 calls to triple zero successfully connected, while 604 experienced an error.

Telstra also operates the triple-zero platform for all telcos, and the company said the platform does not use the NTP servers for synchronisation, so it was unaffected by the incident. Fixed-line callers on the NBN were also not affected.

Telstra has promised to compensate customers hit by the outage, and Brady said the company was working through how that would look for business customers, including the Australian Rail Track Corporation – which operated the network that brought V/Line services to a halt.

Brady said she expected most customers would be repaid in credit. As of Friday, Telstra had received about 8,000 claims from the nearly 9 million customers who had been affected. About $100,000 had been paid out in the past week.

The chair of the committee, Greens senator Sarah Hanson-Young, observed Telstra had been “pretty smug” in earlier hearings in the inquiry – which was originally focused on last year’s Optus outage. Hanson-Young characterised Telstra’s position as an “Optus problem”.

“Well, I’m sorry, today we see it’s not an Optus problem. It’s also a Telstra problem,” she said.

Hanson-Young pointed to Telstra’s reported 3,641 outages in 2024 and 5,221 in 2025, while banking a 2025 profit of $2.3 bn – a 31% increase on the previous financial year.

“So … you’re banking huge increases in profit [while] there are more outages, less reliability for people to access and use their mobile phone,” she said.

“I don’t think it washes to go around telling people that your system is resilient.”

Brady said mobile networks were “complex” and part of an evolving technology environment.

“Certainly networks are not infallible, but our job is to make sure we are taking the actions and we are taking those steps that will mitigate as best we can,” she said.

“I would love to be able to sit here and say there will be zero outages – the reality of a complex network environment with fast evolving technology, you just can’t.

“No telco around the world could guarantee that.”

Brady said the number of minutes a customer’s phone was out was down 91% from the previous year, and the time to restore an outage was down 22% – an improvement of the company’s network resilience, despite the volume of outages going up.