Tencent is considering issuing up to $5 billion in offshore bonds to bankroll its AI push, the Wall Street Journal's Chinese edition (Wallstreetcn) reported on October 9, with East Money and Huxiu carrying the same story a day earlier; HK01, under the headline "raising HK$39 billion offshore again," noted that Tencent's capital expenditure has surged 176% year over year as it goes all-in on AI. If completed, it would be the most direct "compute financing" move by a Chinese tech giant in recent memory.

From operating profit to debt: the changing scale of AI spending

$5 billion is roughly HK$39 billion — not small change even for Tencent. What stands out is the funding channel: Big Tech used to absorb AI spending from operating cash flow; turning to the offshore bond market signals that the scale and certainty of AI capex have outgrown cash-flow comfort zones. If the bond is fully issued, the proceeds point almost entirely at AI infrastructure.

The financing side of the global compute arms race

· Tencent: weighing up to $5 billion in offshore bonds to scale up AI investment (Wallstreetcn, East Money, Huxiu)

· DeepSeek: new round of at least $12 billion, with Tencent and CATL as the largest backers (Bloomberg, Reuters; our October 7 report)

· SoftBank: reportedly seeking funding from Gulf states to double down on AI (Kyodo News)

· SpaceX: reportedly borrowing from Wall Street to buy GPUs (Yeeyi, single source, for reference only)

Why everyone is "locking in" compute

Compute is shifting from a procurement question to a finance question. Under model-iteration competition, the supply certainty of GPUs and data centers directly sets product rhythm: training schedules, inference costs and fleet-scale agent deployment all sit on the compute ledger. Debt financing brings duration matching — AI infrastructure lives for years, and funding it with long-dated bonds is more sustainable than squeezing current-period profits. That is why players from Musk's Terafab fabs to SoftBank's Gulf capital are opening the entire financing toolbox.

When compute becomes a strategic asset, the balance sheet is the new compute scheduler: whoever locks in supply at lower cost and longer duration buys that much more calm for model iteration.

For China's AI industry, Tencent's move carries an extra signal: after DeepSeek's mega-round and Moonshot's pre-IPO financing, a platform giant going long on infrastructure means the compute race has escalated from startup level to platform level. For enterprise users, this capex wave should eventually translate into more inference supply and lower per-call costs — at NineZenith, when helping enterprises plan AI compute and agent resource scheduling, we likewise put multi-source redundancy and cost elasticity first. (Compiled from Wallstreetcn, East Money, Huxiu, HK01, Kyodo News and other public reports)