When objects of unknown origin struck an Amazon Web Services data centre in the United Arab Emirates on Sunday, knocking out an entire availability zone and triggering a fire that required the local fire department to cut all facility power, the incident crystallised something that risk managers have long warned about: critical digital infrastructure sits inside geopolitical reality, not above it.
AWS confirmed on its AWS Health Dashboard that the mec1-az2 availability zone within its ME-CENTRAL-1 region was taken offline after the objects "created sparks and fire." The company offered no explanation of what the objects were or where they originated, but the incident unfolded as the UAE faced Iranian retaliatory strikes following US and Israeli military action. Multiple reports confirmed explosions across the Emirates in the hours surrounding the outage. The other availability zones in the ME-CENTRAL-1 region continued operating normally, and by early afternoon PST, AWS reported positive signs of recovery for EC2 services, with a resolution expected within two to three hours.
The episode raises hard questions about the wisdom of concentrating hyperscale cloud capacity in a region that is now, indisputably, an active theatre of conflict. AWS is not alone in its exposure: as Data Center Dynamics has reported, Microsoft has committed approximately $15.2 billion to UAE cloud and AI infrastructure. When debris, whether from a drone, a missile fragment, or something else entirely, can bring down an availability zone serving thousands of businesses across the Middle East, the assumption that cloud redundancy insulates customers from physical-world risk looks rather less robust than the sales brochures suggest.
Meanwhile, the technology story that carried more immediate consequence for Australian workers broke not in the Gulf but in a regulatory filing lodged with the Australian Securities Exchange. Sydney-based supply chain software vendor WiseTech Global announced on 25 February that it intends to eliminate approximately 2,000 positions over the next two years, representing nearly 30 per cent of its global workforce across 40 countries. CEO Zubin Appoo did not dress it up: "The era of manually writing code as the core act of engineering is over," he wrote in the filing.
Appoo told analysts that tasks previously requiring six or seven months can now be completed in a single day, and that rolling out customs capability in a new country, once a two-year undertaking, can now be done six or seven times faster thanks to AI. Product and development teams will face the steepest cuts, with some divisions potentially losing half their headcount. WiseTech's US cloud arm, E2open, acquired in August for $2.1 billion, may see reductions of up to 50 per cent. Around 500 roles had already been eliminated since last July before the announcement was made.
The market's reaction was telling. WiseTech shares jumped as much as 11 per cent in Sydney trading on the day of the announcement, even as the stock remained well below its November 2024 peak. Investors, it seems, read the cuts as a sign of fiscal discipline rather than distress. The company simultaneously reported a first-half underlying net profit of $114.5 million, six per cent ahead of market consensus, and reaffirmed its full-year outlook. Critically, WiseTech has also restructured its revenue model: approximately 95 per cent of CargoWise customers have already moved to a transaction-based pricing model, insulating the company from the revenue erosion that seat-based licensing faces as AI reduces headcount at client organisations.
That strategic repositioning deserves scrutiny alongside the sympathetic hearing it received from investors. WiseTech's CargoWise platform processes an estimated 75 per cent of worldwide customs transaction data and is used by 24 of the 25 largest global freight forwarders. When a company with that kind of systemic reach in logistics declares manual coding obsolete and restructures accordingly, the effects ripple well beyond its own payroll. The workers losing roles are real people, and the Fair Work Commission and Australian unions will be watching how the company manages its redundancy obligations across multiple jurisdictions.
Those who argue that AI-driven efficiency gains ultimately create more jobs than they destroy have a reasonable historical case. Automation has repeatedly transformed labour markets without permanently destroying aggregate employment. What is genuinely harder to assess is the transition cost borne by individual workers in the interim, particularly in specialised technical roles where retraining is neither quick nor cheap. The honest position is that both things can be true: WiseTech may well be more competitive and productive in three years, and the 2,000 people displaced will still face real hardship in the meantime.
Factories Without Workers, Chips Without Fabs
WiseTech's announcement did not arrive in isolation. Samsung Electronics separately declared its intention to operate fully autonomous factories by 2030, according to a company statement outlining plans to integrate AI agents across its entire manufacturing value chain, from inbound logistics through to quality inspection and final shipment. The South Korean giant said it would progressively introduce humanoid and task-specialised robotics across its global production lines, covering line operations, materials handling, and precision assembly. The framing was deliberate: Samsung described the shift as a "transition from automation to advanced autonomy," signalling that this is not incremental improvement but a structural redesign of how things are made.
On the supply side of the semiconductor industry, Micron Technology formally opened its first plant in India, a semiconductor assembly and test facility in Sanand, Gujarat. Micron originally announced the investment in 2023, and while the facility is not yet fully completed, the company said the first phase will feature more than 500,000 square feet of cleanroom space once fully ramped, making it one of the world's largest single-floor assembly and test cleanrooms. India's government has been aggressively courting semiconductor investment, and Micron's presence adds credibility to New Delhi's ambition to become a meaningful node in the global chip supply chain.
Indonesia is pursuing a different entry point to the same race. The government last week signed a strategic partnership with UK chip design house Arm under which Indonesia will attempt to train 15,000 engineers on the Arm ecosystem. Some engineers will be sent abroad; Arm trainers will also visit Indonesia directly. The government plans to select six priority industries to develop chip expertise, positioning the country to participate in semiconductor design rather than purely in manufacturing assembly.
Taken together, these stories from a single week form a coherent if uncomfortable picture. Digital infrastructure is physically vulnerable in ways that cloud marketing rarely acknowledges. AI is eliminating technical jobs faster than most workers or policymakers anticipated. And the race to build sovereign semiconductor capacity, whether in India, Indonesia, or Australia's own nascent advanced manufacturing ambitions, is accelerating precisely because governments have concluded that dependence on a handful of concentrated supply chains is a strategic liability. Reasonable people can disagree about the right policy responses to all of these pressures. What is harder to dispute is that the decisions being made in corporate filings and factory announcements this week will shape Australian economic life for a generation.