Australian cable firm Vocus has joined the chorus of those warning about a subsea capacity crunch. Simon Parker, Vocus’ head of strategic sales, said the company was seeing a “tidal wave of capacity demand” and the industry was running short of stock.
“That applies not just to lit capacity, but more critically for us, to the underlying fiber that actually carries that capacity,” Parker told an online event last week. “Our customers aren’t just asking for double or quadruple of what they had. Some of them are asking for 100x uplifts in capacity.”
Vocus operates domestic subsea and terrestrial capacity in Australia as well as a major route to Singapore, as well as a share in a Google cable across the Pacific.
Parker said one response to skyrocketing demand was to maximize the capacity of cables in the water, which meant close collaboration with partners in the supply chain. The other answer is to build fresh capacity, but most of the construction resources have been captured by the hyperscalers.
Extremely short supply
“We need cable manufacturing slots, marine installation vessels, and permitting. We need specialist people who run these projects. These are all items in extremely short supply because the hyperscalers are consuming those resources themselves.”
These resources are also heavily committed to beyond 2030, which means the only alternative is to work closely with US giants who are dominating the buildouts, Parker said.
He admits it’s an approach others in the industry reject, because of concerns that reliance on the foreign heavyweights risks surrendering control over domestic and regional infrastructure. He says that Vocus also holds some of those concerns, but argues the sovereign investment that people advocate as an alternative is just not there.
Hyperscalers and neoclouds “are the only parties with the financial resources to build purpose-built infrastructure for the AI ecosystem, and the scale and the speed that this moment actually needs,” he said.
Parker is also one person who doesn’t fear an AI bubble. He says that unlike the dotcom era, the hyperscalers driving AI investment have real balance sheets behind them and are meeting genuine demand.
He notes that while those companies are forecasting $760 billion in capex investment this year, in Q2 alone they generated $660 billion in operating cash flow – up 33% over last year.
