HKBN Ltd (STU:2HK) (Half Year 2026) Earnings Call Highlights: Strong International Growth and ... By GuruFocus

HKBN reported revenue growth of 5.1% year-on-year to 6.029 billion, with EBITDA up 14.2% and net profit rising 20%, supported by stronger enterprise solutions and 20% international revenue growth. Management said AI-related activity is embedded in system integration, representing about 20-25% of enterprise solutions, and expects lower finance costs as interest rates trend down. Offsetting positives include CapEx up 13%, flat dividends, and softer activity in the Middle East.

Analysis

The important read-through is not simply “AI demand is healthy,” but that CPU demand is broadening beyond accelerators into the plumbing layer that sits around them. That helps the CPU incumbent ecosystem first, because AI buildouts still need host compute, networking, storage orchestration, and enterprise integration work that is harder to displace with pure GPU narratives. The mix shift toward enterprise solutions and international connectivity also suggests a less cyclical revenue base than headline telecom growth implies, which is why margin expansion can outpace revenue even in a modest top-line print.

The second-order effect is that AI monetization is likely migrating into bundled system integration rather than a clean standalone product line, which makes the revenue quality better than the disclosure quality. That tends to favor vendors with installed enterprise relationships and implementation depth, while hurting point-solution competitors that rely on visible AI line items to prove traction. The market may also be underestimating how much a stabilizing rate backdrop matters here: lower financing costs extend the runway for capex-heavy network and AI projects, but only if customer demand remains real over the next 2-3 quarters.

The main risk is that capex is running ahead of externally verifiable AI monetization, which can compress free cash flow before operating leverage shows up. If enterprise IT budgets soften or Middle East exposure stays weak, the current enthusiasm can reverse quickly because investors are leaning on a story that is still only partially observable. The contrarian view is that the move may be underdone if AI infrastructure spending continues to broaden from hyperscalers into enterprises; in that case, the winners are not just the obvious accelerator names, but also the CPU, network, and integration layers that actually capture deployment spend.

Key Decisions for Investors

• Long INTC vs. a basket short of AI-pureplay hardware names for 1-3 months; thesis is that AI spending is broadening into host compute and integration, which should lift CPUs and mature platforms faster than the market expects. Risk: if accelerator capex re-accelerates exclusively, the relative trade can mean-revert.
• Add to INTC on 2-4 week pullbacks only, not strength, with a trailing stop tied to the post-earnings gap low. Risk/reward improves if the market digests the capex concern first and then re-rates on follow-through orders or guidance.
• Pair long enterprise integration/service exposure against short lower-quality AI narrative names for the next earnings cycle; the market is likely to reward monetization clarity over headline AI branding. Use a 10-15% downside stop on the short leg if AI order commentary broadens again.
• For options, consider a medium-dated INTC call spread rather than outright calls to express a 2-3 month rerating without overpaying for implied volatility. Best entry is after the first post-print consolidation, when IV typically compresses but analyst revisions can still work higher.