On August 14, CITIC Securities International released a research report that should have set off alarm bells for any investor trained in quantitative risk asymmetry. Tencent’s second-quarter performance slightly exceeded expectations, driven by domestic gaming and advertising. But the real story lies in the capital expenditure forecast: HKD 215.7 billion for 2026, HKD 260 billion for 2027. The report simultaneously lowered core net profit estimates for 2026-2028 by 5% to 9%. This is not a bullish signal. It is a structural imbalance between promise and payoff.
Let me be clear: Tencent is not a blockchain protocol. But the capital allocation dynamics here mirror exactly what I observed during the 2020 DeFi yield trap. High yield is a warning, not a welcome. In DeFi, inflated annual percentage yields (APYs) temptation investors while the underlying liquidity pools bleed. Here, inflated capital expenditure forecasts promise AI-driven growth while depreciation costs quietly erode net profit. The pattern is identical: a promise of future returns that cannot be sustained by the current financial architecture.
Context: The Illusion of the Moat
Tencent has clarified four major AI strategies, supporting more aggressive capital expenditure. The logic is straightforward: core business profitability (gaming, advertising, WeChat ecosystem) provides the cash flow to fund AI investments. The report argues that this profitability release supports the AI spend, and that AI investments have clear downside protection because of the core business strength. This is the same argument made by Terra’s founders in 2022: the algorithmic stablecoin’s value was “protected” by the Luna burn mechanism. Code does not lie; people do. The downside protection in Tencent’s case is not a financial safeguard—it is a narrative one.
CITIC Securities maintained a “Buy” rating, adjusting the target price from HKD 632 to HKD 620. A 1.9% target price reduction while raising capital expenditure by billions? That is a mathematical contradiction. The report’s own numbers reveal the tension: they expect core net profit growth rates of only 2% and 3% for 2026 and 2027, respectively. Capital expenditure is growing at a compound annual growth rate (CAGR) of over 20% (from HKD 215.7B to HKD 260B in one year). The gap between investment growth and profit growth is a structural risk. In the crypto world, we call this a “rug pull” of valuation.
Core: Systematic Teardown of the Capital Expenditure Plan
Let me dissect this using the same forensic methodology I applied to the 0x v2 integer overflow vulnerability in 2018. The vulnerability was hidden in the maker fee calculation logic. Here, the vulnerability is hidden in the depreciation schedule.
First, the raw numbers. Tencent’s 2026 capital expenditure is forecast at HKD 215.7 billion. Assuming a conservative 5-year straight-line depreciation for AI infrastructure (servers, GPUs, data centers), the annual depreciation expense would be approximately HKD 43.1 billion. For 2027, with HKD 260 billion, the depreciation adds another HKD 52 billion annually. Cumulatively, by 2028, the total depreciation charge from these two years alone could exceed HKD 200 billion per year. Core net profit growth is projected at 2-3% year-over-year. That means the depreciation expense will consume any incremental profit growth, and likely more.
CITIC Securities acknowledges this: “due to rising depreciation costs, they have lowered the core net profit estimates for 2026 to 2028 by 5% to 9%.” But they frame this as a temporary pressure that can be alleviated by core business profit improvement. This is a logical fallacy. If core business profit improves by 2% but depreciation costs increase by 20%, the net effect is negative. The only way this works is if the AI investments generate incremental revenue that exceeds the depreciation. But the report does not provide such revenue forecasts. This is a gap I call the “revenue linkage failure.”
During my 2024 Bitcoin ETF structural critique, I identified a similar gap: the custody solutions of major issuers had conflicts of interest that were not disclosed. Here, the conflict is between the narrative of AI growth and the reality of financial statements. The report’s own data shows that operating profit (excluding AI investment) increased by 19% year-on-year. But the AI investment is not excluded from the capital expenditure. The operating profit growth is being used to justify the capex, but the capex will eventually depress operating profit depreciation. This is a classic example of using a one-time positive signal to justify a multi-year negative structural change.
Let me run a sensitivity analysis. Assume Tencent’s core business operating profit in 2025 is HKD 200 billion (a rough estimate based on mid-cycle earnings). With a 2% growth rate, 2026 operating profit would be HKD 204 billion. Depreciation from the new AI capex adds HKD 43 billion, assuming no other changes. That would reduce operating profit to HKD 161 billion, a 19.5% decline. The report’s lower core net profit estimates of 5-9% are optimistic. If the AI capex continues to grow, the depreciation drag will compound.
This is the same death spiral I reconstructed in the Terra/Luna collapse. The Luna burn mechanism created a feedback loop: as demand for UST fell, Luna was burned, reducing supply, but the lack of external collateral meant the burn accelerated the price decline. Here, the AI capex creates a feedback loop: as depreciation increases, net profit falls, which reduces the cash available for future capex, forcing either debt or equity dilution. The “downside protection” cited by the report is a fiction.
Contrarian: What the Bulls Got Right
But I am not a complete nihilist. The bulls have a point: Tencent’s ecosystem moat is real. WeChat, gaming, and advertising generate stable cash flows. The AI investments are not speculative gambles on a new token; they are infrastructure for an existing user base. Tencent’s AI products (e.g., Hunyuan large model, cloud AI services) have a distribution channel that most crypto projects lack. In the 2026 AI-agent crypto integration audit I conducted, I found that most projects failed because they lacked a user base. Tencent does not have that problem.
Furthermore, the report’s capital expenditure forecasts may be conservative. If AI adoption accelerates, the revenue from AI services could offset the depreciation. Tencent could also monetize the AI infrastructure through cloud services, which have higher margins than gaming. The report’s 2% profit growth estimate may be a floor, not a ceiling.
But this is where the logic breaks. The report raised capex forecasts for 2026 and 2027 but lowered profit estimates. That means the analysts themselves do not believe the revenue from AI will be sufficient to cover the depreciation in the short term. They are essentially saying: “Invest now, profits later.” This is the same narrative that drove the 2020 DeFi yield trap. High yield is a warning, not a welcome. The warning here is that the depreciation schedule is a liability that will crystallize within two years.
Takeaway: Accountability Call
Tencent’s AI capital expenditure is a structural risk that the market is currently pricing as a positive. The depreciation costs will eat into net profit, and the promised revenue linkage is absent. Investors should ask: “What is the expected return on this HKD 260 billion? Show me the revenue model, not just the cost breakdown.”
Forensics don’t lie. The numbers are clear: a 20% capex growth rate paired with a 2% profit growth rate is unsustainable. Tencent will either have to slow the capex, issue debt, or accept lower margins. The market’s “Buy” rating is based on the assumption that the AI investment will eventually pay off. But in the crypto world, we have learned that assumptions are liabilities. Code does not lie; people do. Audit the promise, not the poster.
The question is not whether Tencent’s AI strategy is viable. The question is whether the current capital allocation is mispriced. Based on the data, the answer is yes. The depreciation drag is a ticking clock. Investors should demand a revenue forecast, not just a cost projection. Until then, treat this as a high-yield warning: the promise of future returns is not backed by current financial reality.
High yield is a warning, not a welcome. Tencent’s capital expenditure is no exception.