CEVA Logistics has finished the first six-month phase of moving heavy equipment for AirTrunk’s hyperscale data centre campus in Western Sydney, and is now running the same operation in Johor Bahru, Malaysia. The work covers end-to-end logistics: warehousing, customs clearance, domestic transport, and project management for kit including 45-tonne generators and oversized cooling towers. The contract signals how physical delivery is shaping the pace of AI infrastructure across Australia and Southeast Asia.
AirTrunk says the logistics work improved the safe movement of critical gear during construction of the Sydney campus. The harder question is who controls the speed of the build, and what that decides downstream.
Building an AI data centre is starting to look less like a computing problem and more like a freight one. The chips and the cooling are solved on paper. Getting a 45-tonne generator onto a construction site in Western Sydney, on schedule, without damage, is not.
That gap is what CEVA Logistics is now filling for AirTrunk, the Australian developer of hyperscale data centres for cloud and AI customers across Asia-Pacific. CEVA has completed the opening six-month phase of AirTrunk’s Sydney campus and shifted the same playbook to Johor Bahru, Malaysia. The job runs end to end: warehousing, customs, ground transport, and the sequencing that keeps a build from stalling.
It reads like a routine contract announcement. It is not. The real contest underneath is which operators can pour AI capacity into the region fastest — and the bottleneck has quietly moved from silicon to logistics.
The bottleneck moved from silicon to the loading dock
The equipment list tells the story. A hyperscale data centre needs dense power and heavy cooling, which means generators, transformers, and cooling towers that are too large for ordinary freight handling. CEVA is moving and storing this kit using a network that combines contract logistics with air, ocean, and ground freight. That integration is the point: one operator controls the whole chain rather than handing off at each border.
Guy Meredith, Vice President of Contract Logistics APAC at CEVA, says the company is seeing supply-chain demand climb sharply as AI-driven data centre construction speeds up across the region. AirTrunk, for its part, credits the logistics work with keeping critical equipment moving safely through the Western Sydney build.
Here is the forward read. Once a developer secures power, land, and permits, those constraints are fixed for years. The variable that still moves a launch date by months is whether the heavy kit arrives in the right order. Over the next twelve to eighteen months, the firms that compress that handling risk will set the pace of AI capacity in Australia and Malaysia — not the firms that simply own the most servers.
The honest limit: this is one contract, and CEVA’s own framing drives much of the public detail. What it does not show is hard capacity numbers or completion dates for either site, which remain unconfirmed. The mechanism is real. The scoreboard is not yet public.
The physical backbone of the AI race is being poured now
The contest here is between regional specialists like AirTrunk and the global hyperscale field led by U.S. cloud platforms and large colocation firms. AirTrunk’s edge is regional scale and the speed at which it develops sites across Australia and Southeast Asia. Its constraint is the same one everyone faces: power, land, and permits.
Winning this race means owning the physical backbone for AI workloads and latency-sensitive cloud services. Whoever controls that capacity in the region controls when global digital services get the compute they need.
For Western firms, the exposure is indirect but real. A company running cloud workloads may have them land in Australia or Malaysia without ever choosing the location directly. The practical risk is jurisdictional fragmentation — different storage, transfer, and disclosure rules applying depending on where the workload physically sits, a tension already visible in how China and Malaysia are negotiating cross-border data rules.
So the routine logistics contract turns out to be a marker in a larger fight. The chips get the headlines. The freight schedule decides who gets there first.
Beyond the headline
The bigger picture
The deeper shift is that AI infrastructure is becoming a logistics problem as much as a computing one. Once power, land, and permits are secured, the bottleneck moves to moving heavy kit, sequencing deliveries, and keeping build schedules intact across multiple countries.
The money trail
The beneficiary is not just the data-centre developer but the ecosystem that turns construction speed into revenue: logistics contractors, equipment suppliers, and power-infrastructure vendors. In a capital-heavy buildout, the firms that cut delay and handling risk can capture more value than the companies that only own the servers.
The reach
CEVA is the actor, cross-border project logistics is the mechanism, and the non-obvious result is faster regional rollout of AI capacity for global cloud users. That matters because physical delivery schedules now shape when large-scale compute becomes available to Western-facing digital services.
What the buildout means for your workloads
As AI capacity concentrates in Australia and Malaysia, two groups need to track where their data and money actually sit.
- Businesses routing cloud workloads through APAC
If your vendors host sensitive data on Australian infrastructure, check the Office of the Australian Information Commissioner guidance on cross-border disclosures before you sign. For storage in Johor or elsewhere in Malaysia, review the Department of Personal Data Protection guidance on processor obligations — the rules differ by jurisdiction, and they apply wherever the workload physically lands.
- Investors tracking AI infrastructure
The value in this buildout is not only in the server owners. Watch logistics contractors, generator and cooling suppliers, and power vendors over the next 12 to 18 months. The firms that reduce build delay capture margin that pure compute providers cannot.
Explainer
- Hyperscale data centre
- A very large facility built to house thousands of servers and networking systems for cloud and AI work. Its defining traits are dense power delivery, heavy cooling, backup redundancy, and fast internal connections that ordinary server rooms cannot match. AirTrunk builds these across Asia-Pacific and the Middle East, where site development speed has become a competitive advantage in its own right.
- AirTrunk
- An Australian company that develops hyperscale data centres for cloud, AI, and data-intensive customers. It operates across Asia-Pacific and the Middle East, competing on regional scale and the speed at which it secures and builds sites. Its current Western Sydney and Johor Bahru projects sit at the centre of the region’s race to add AI compute capacity.
- CEVA Logistics
- A global logistics and supply-chain company offering integrated freight and contract logistics services. Its network combines warehousing and ground transport with air and ocean freight, allowing one operator to manage cross-border deliveries end to end. For data centre construction, that integration matters because it removes the handoffs where oversized cargo most often gets delayed or damaged.
RELATED STORIES
