Tencent’s ‘Midway Moment’: AI Strategy Shifts Market Perception, BofA Sees 77% Upside
TREE NEWS reports: In a significant development for the global AI race, Tencent is emerging from its status as a perennial catch-up player. Following a post-earnings investor roadshow in Hong Kong, BofA Securities has declared that the company is at a ‘Midway battle’—a turning point where the balance of power in AI is subtly shifting. The bank maintained a ‘Buy’ rating and a HK$780 target price, implying over 77% upside from the current share price of HK$440.
The roadshow, which featured Tencent’s Chief Strategy Officer James Mitchell, CFO John Lo, and IR team, was dominated by investor questions on AI strategy—capex outlook, monetization paths, compute market dynamics, and model progress. BofA analysts Alex Liu and Joanna Du highlighted that Tencent’s rapid catch-up in model capabilities and compute accumulation marks a potential inflection point in the long-term competitive landscape.
Compute as Strategic Asset, Not Just Cost
Management emphasized that compute ownership is a strategic asset, supporting not only cloud monetization but also foundational model competitiveness and the broader ecosystem. Tencent significantly boosted AI infrastructure capex in Q2 2026 and made substantial prepayments to lock in memory and compute supply before price increases. The bank raised its 2026/2027 capex forecasts to RMB 210 billion and RMB 260 billion, respectively, from RMB 185 billion and RMB 225 billion. However, quarterly cash pressure from compute investments is expected to peak in Q2-Q3 2026 and normalize by 2027.
Model Strategy: Efficiency Over Benchmark-Chasing
Tencent is not chasing benchmark rankings. Instead, it focuses on improving model capabilities while maintaining attractive inference cost efficiency. The Hunyuan model is set for faster iteration, with the HY4 version expected within three months, narrowing the gap with larger domestic models. The WeLM model, powering the WeChat AI assistant ‘Xiaowei’, is designed for high cost efficiency, requiring fewer compute resources and compatible with a wider range of GPUs.
AI Applications: WorkBuddy and Xiaowei
WorkBuddy is positioned as a key AI application opportunity, benefiting from a positive feedback loop—real user interactions improve model intelligence, enhancing product performance. Monetization is early, but there is significant room for user growth and feature expansion. Xiaowei is being rolled out cautiously, with potential applications in travel planning, price comparison, and daily productivity. Full-scale rollout is possible as early as Q4 2026, subject to partner progress.
Earnings Pressure from Depreciation
Aggressive AI investment brings near-term earnings pressure. The bank cut 2026-2028 non-IFRS net profit forecasts by 2-6% due to accelerated depreciation. It now expects 2027 non-IFRS net profit to be flat year-over-year. However, cloud revenue growth is expected to accelerate in H2 2026 and 2027, surpassing the ~20% YoY growth in Q2 2026.
Key Takeaways for Investors
- Valuation disconnect: At under 13x 2026/2027 P/E, the market is not pricing in Tencent’s AI option value (token and cloud revenue upside) or core business resilience.
- Catalysts: Three near-term catalysts: Hunyuan model upgrade (3 months), Xiaowei full rollout (6 months), and WorkBuddy/cloud revenue acceleration (12 months).
- Risk: Depreciation will pressure near-term earnings; investors must balance long-term AI upside against short-term profit dilution.
- Strategic shift: Tencent’s compute flexibility and cost-efficient models position it as a formidable player, not just a follower.
In summary, the ‘Midway battle’ metaphor is apt: the war is not over, but the momentum is shifting. For investors, this could be an opportune moment to reassess Tencent’s role in the AI revolution.
