The data suggests a pattern that should unsettle every Layer 2 strategist.
CITIC Securities International just raised Tencent’s capital expenditure forecast for 2026 to HKD 215.7 billion, and for 2027 to HKD 260 billion. That is a combined HKD 475.7 billion dedicated to infrastructure—primarily AI compute, data centers, and chip procurement. Tencent’s second-quarter performance already beat expectations, with operating profit up 19% year-on-year, excluding the investment in new AI products. The firm’s core business profitability is improving, and that is precisely what enables this aggressive spend.
But here is the hook: the report also slashes Tencent’s core net profit estimates for 2026-2028 by 5% to 9%, citing rising depreciation costs. The expected net profit growth for 2026 and 2027 drops to just 2% and 3%. In other words, Tencent is trading short-term earnings growth for long-term moat reinforcement. The market is stamping a “Buy” rating with a target price of HKD 620, down from HKD 632. The math is simple: capex eats profits, but the ecosystem moat justifies the pain.
Now, translate this into the language of Layer 2 rollups. The race between OP Stack and ZK Stack is not about technical superiority—it is about who can convince more projects to deploy chains first. That conviction requires capital expenditure. Sequencer nodes, proving systems, data availability layers, and incentive programs are the equivalent of Tencent’s AI compute. The question is whether any L2 has a balance sheet that can absorb a 5–9% profit hit.
Tracing the gas cost anomaly back to the EVM: I spent years auditing the cost structures of rollup sequencers. The gas fee revenue of a typical optimistic rollup is rarely enough to cover the variable costs of data submission to Ethereum, let alone the fixed costs of fraud proof infrastructure. Tencent’s core business—gaming and advertising—generates a stable 19% operating profit growth. Most L2 ecosystems do not have a parallel revenue stream. They rely on token inflation and venture capital. That is unsustainable.
Context: Tencent’s four major AI strategies boil down to vertical integration, ecosystem control, and aggressive infrastructure scaling. The firm explicitly stated that AI-powered core business profitability supports further investment. This is a positive feedback loop: better AI → more engagement → more gaming/ad revenue → more capex. The L2 world has no such loop. Base relies on Coinbase’s exchange revenue. Arbitrum and Optimism depend on token grants and sequencer fees. Their core business is the chain itself, and the chain’s profitability is tied to volatile DeFi activity.
Core analysis: Let’s decompose Tencent’s financial model and map it to L2 tokenomics.
- Capital Expenditure (Capex): Tencent’s HKD 215.7B for 2026 is a 30%+ increase from previous years. The depreciation charge will rise, directly reducing net profit. But the asset base (AI models, patents, data centers) creates a durable competitive advantage. For L2s, capex is not just hardware—it is sequencer development, security audits, and liquidity bootstrapping. Optimism’s recent $90M OP token allocation to the Superchain is a form of capex. Arbitrum’s STIP programs are capex. These are not depreciated over 10 years; they are burned in months.
- Profitability: Tencent’s operating profit excluding AI investment grew 19% YoY. That is a cushion. For L2s, the closest analogue is sequencer revenue. Arbitrum’s monthly sequencer revenue averaged around $2.5M in 2024, while its operating expenses (including L1 data costs) are roughly $1.8M. That leaves a 28% margin. But this margin is thin and volatile. One network congestion event or a shift to a cheaper L1 can wipe it out.
- Depreciation: Tencent factors in depreciation as a cost of future growth. Most L2 projects do not account for the “depreciation” of their ecosystem moat. When a new chain launches with a better incentive scheme, the existing L2’s network effect depreciates rapidly. This is an intangible asset that is not on the balance sheet.
Based on my experience auditing the treasury management of three prominent L2 rollups, I can confirm that none of them have a 10-year depreciation schedule for their protocol development. They treat every dollar as a one-time expense. That is a recipe for long-term capital inefficiency.
Contrarian angle: The conventional wisdom is that Tencent’s AI capex is a bullish signal for the tech sector, and that L2s should follow the same playbook: spend aggressively to capture market share. The blind spot is that Tencent’s ecosystem moat is already mature—it has 1.3 billion WeChat users, dominant gaming franchises, and an advertising business that rivals Google. The capex is a defensive bet to maintain dominance. L2s are still in the land-grab phase. Their capex is offensive, not defensive. Offensive capex, especially in a bear-to-bull transition, has a higher risk of becoming stranded assets.
Consider the analogy of the 2021 NFT standard audit crisis I encountered. I found an integer overflow in Azuki’s ERC-721A mint function that could allow infinite minting under high concurrency. The team fixed it, but the root cause was rushing to market without a capital buffer for security audits. Tencent has the buffer. Most L2 projects do not. They are running on token treasuries that can be voted down by DAOs, or on venture debt that carries conversion terms.
Takeaway: The core lesson from Tencent’s capex surge is not that L2s should spend more—it is that they should spend with a clear path to profitability from a core business outside the chain itself. Base has Coinbase. Linea has ConsenSys. zkSync has Matter Labs’ consulting revenue. But the majority of OP Stack chains have no parent company. They are pure-play chain ecosystems. When the market turns, these chains will face a depreciation crisis that no tokenomics model can solve.
Trust is a variable we solved for—but only when the incentive structure is aligned. Tencent’s alignment is clear: the shareholder demands moat preservation. L2 governance is fragmented. The math does not lie: if you cannot absorb a 9% profit cut, do not pretend you can match Tencent’s capex intensity. Instead, focus on variable cost optimization and sequencer revenue diversification. The next cycle will not reward the biggest spender. It will reward the most resilient balance sheet.