Recently, the tech world has been a bit influenced by the "money smell."

According to reports, Tencent is considering issuing up to $5 billion in offshore bonds. This is not just Tencent alone "raising money"; Alibaba, ByteDance, and DeepSeek have also recently been intensive fundraising. The four giants together hold nearly 400 billion RMB in cash.

You might ask: aren't these companies the most profitable? Why are you in a hurry to borrow money compared to us?

The answer is simple: they are racing against time. During the window period when funding costs are still low, secure the "AI entry ticket" for the next few years in advance. This money wasn't for salaries, but for buying "shovels" and "foundations."

| Tencent's Plan: Borrowing Money Is Just to "Move Around"

Tencent's logic behind this bond issuance is clear: using debt to buy space.

In June this year, Tencent's capital expenditure surged 176% to 52.8 billion yuan, and free cash flow even turned negative. Where did the money go? All of it is poured into AI projects and computing power construction. At the same time, Tencent still has to maintain annual stock buybacks worth hundreds of billions to appease shareholders.

Having to develop AI, buy backs, and invest, cash flow is inevitably tight. Issuing bonds at this time is the optimal solution.

Timing mismatch: AI requires long-term investment and may not recoup in the short term. Using long-term bonds to cover long-term investments is more manageable than spending cash on hand.

Debt structure: Tencent has no public bonds maturing this year; the next large amount maturing is in 2028. Borrowing money now is like postponing repayment pressure, giving yourself plenty of time to "develop."

| A fierce competition of the four giants' "money-making" strategies

Although all are engaged in AI, the four giants each have their own unique financing methods, making them textbook examples:

腾讯（发债）：稳扎稳打。利用高信用评级借低息美元/人民币，主打一个“手中有粮，心中不慌”。

ByteDance (syndicated loan): Budgeting carefully. It borrowed $29.6 billion, and the spread is 17 basis points lower than in 2024. Twenty-eight banks rushed to lend money to it, showing their strong confidence in ByteDance's profitability.

Alibaba (Rights Issue): Equity financing. issued new shares totaling HKD 80 billion. Although equity dilution is possible, equity funds do not require principal or interest payments, making them ideal for infrastructure projects with long payback cycles (such as global computing centers).

DeepSeek (primary market financing): valuations soaring. From 50 billion yuan to nearly 100 billion yuan, Tencent and CATL are rushing to invest. This is a classic case of "burning money for the future," betting on the endgame of AI large models.

|Where did the money go?

Is this 400 billion spent on the market to make a splash? No, most of this money is spent on hardware and infrastructure.

DeepSeek: Currently building a 1GW data center in Ulanqab, Inner Mongolia, it spent $2.56 billion just to buy Huawei's Ascend 950DT chips.

Upstream Meat: Wherever the money flows, the grass grows most lushly. Yangtze Memory and Changxin Technology achieved an astonishing 78% and 84% gross margins, respectively. What does this indicate? This shows that selling "shovels" (chips, storage) is more profitable than mining applications (mining applications).

Global resonance: It's not just China—SoftBank has also issued $11.1 billion in bonds to pursue AI. The whole world is leveraging and betting that AI is the next industrial revolution.

Kingtech Perspective | What signals does this "tech giant financing wave" send?

AI has entered the "bayonet battle" phase: previously storytelling, now it is infrastructure competition. Whoever can be the first to build a large-scale computing power cluster will have pricing power.

Upstream Certainty is the Highest: No matter which large model downstream wins, it must buy chips, storage, and build data centers. Upstream hardware manufacturers have the highest earnings certainty.

Pay attention to "water sellers": besides chips, supporting infrastructure such as power, heat dissipation, and optical modules is also a key focus for capital flows.

Short term: Focus on performance releases from upstream hardware suppliers (such as storage and server foundry).

Long-term: Bullish on internet giants with massive cash flows that can quickly build computing power barriers through bond issuance or rights issues. They are turning AI into the new "water, electricity, and coal."