Amidst the AI boom, computing power has emerged as one of the hottest sectors in global capital markets over the past few years.

From initial GPU concepts and computing power leasing to today's smart computing center construction and cloud-based computing services, the investment frenzy across the industrial chain continues unabated. A growing number of listed companies are joining the fray, with multi-billion or even tens-of-billions-of-yuan computing power orders serving as key catalysts fueling market expectations.

However, as the industry matures, new dynamics are reshaping the computing power landscape.

On one hand, leading tech firms are intensifying their investments in AI infrastructure, driving sustained demand for computing power and creating ample room for industry expansion. On the other hand, some listed companies' computing power businesses remain mired in planning, contract-signing, or even conceptual stages, revealing significant gaps between order volumes and actual delivery, as well as between revenue growth and profitability.

This divergence is now starkly reflected in capital markets. While a single cooperation announcement once triggered sharp stock price rallies, market scrutiny of computing power firms has become far more rigorous: Can orders be fulfilled? Can computing power be reliably delivered? Can investments ultimately translate into sustainable operating profits?

As the computing power industry shifts from competing on plans to competing on execution, companies following different development paths may face a new round of divergence.

[Market Action: From +8% to -86% in Just One Week]

On October 13, TechVantage Cloud (00199.HK) opened at HK$2.68 before plunging in a straight-line drop, hitting a session low of HK$0.34. It closed at HK$0.37, down 86.25%. The stock witnessed intraday volatility exceeding 87%, with trading volume surging to over HK$55 million, a volume ratio of 37.63, a turnover rate of 5.50%, and 2,888 executed trades.

This marked a dramatic reversal from a recent rally. Over the past 52 weeks, the stock had traded between a low of HK$0.27 and a high of HK$2.94—surging nearly tenfold in the past two months. Just a week earlier, on October 9, it had spiked over 8% intraday on news of a "major GPU order win."

In one week, the stock swung from a volume-driven rally to a volume-driven collapse.

[What Happened: A Dense Transformation Timeline]

Reviewing the company's actions over the past two months reveals a frenetic pace: In September, it rebranded from "TechVantage Properties" to "TechVantage Cloud"; on September 24, it announced that its AI infrastructure-related businesses met financial reporting standards, formally becoming an independent reportable operating segment with three divisions: AI infrastructure operations, AI computing power orchestration and cloud services, and model services and application platforms. On September 28, it disclosed an HK$83.616 million acquisition of an 80% stake in Guangzhou Chaoyi Xiyang Technology Co., Ltd., whose subsidiary had received 26 high-performance servers and integrated machines totaling 336 cards, and had signed two computing power scheduling service contracts worth approximately HK$649 million in total, with the first contract already initiating delivery and revenue collection.

However, the other side of the story must also be acknowledged: The company's computing power business only began delivery in June and became an independent segment in September, with profit/loss figures still undisclosed—the mid-year report (as of September 30) will be the first to disclose segment-specific data separately. Currently, the disclosed hardware scale stands at 26 servers and 336 cards. The company retains its property development business as its mainstay. Yingli Securities had previously noted that the market widely categorizes it as a "transformation shell stock" and prices it based on its historical real estate operations.

In other words, while the transformation efforts are real, the business remains in its infancy. The stock's extreme volatility essentially reflects a repricing of the gap between "expectations" and "delivery"—while announcements can be frequent, revenues and profits can only be confirmed transaction by transaction.

[A Simple Truth: Industry is the Foundation, Finance is the Tool]

As the computing power rally evolves, the divide between its first and second phases has become clear: The first half was about storytelling—transformation, rebranding, contract-signing, and frameworks. The second half hinges on two things: tangible delivery and realizable profits.

To gauge a computing power firm's quality, three metrics suffice: delivery scale—how much has been delivered and billed stably; utilization rate—whether built capacity is being activated; and profit quality—whether profits come from core operations or one-time gains.

[Five Hong Kong-Listed Case Studies: Same Sector, Five Approaches]

Using these three metrics as a yardstick, five representative firms in Hong Kong's computing power sector demonstrate five distinct approaches.

GDS Holdings (09698.HK): A veteran third-party IDC provider, operating with a focus on "stability." It has 757MW of orders under "take-or-pay" contracts, signed a record 470MW in new contracts in H1, and has a global data center footprint. Trading at ~1.45x PB, it is among the sector's lowest. While asset-heavy with solid orders, its growth elasticity remains moderate.

SUNeVision (01686.HK): A long-established Hong Kong data center operator, it boasts the sector's most robust financials—58.63% gross margin, 18.73% ROE, ~91% occupancy, and a ~2.15% dividend yield. Its drawback is a conservative approach: utilization rates at its MEGA IDC Phase 1 are expected to rise from ~30% to ~70% only after full leasing, with a measured expansion pace.

Kingsoft Cloud (03896.HK): An established cloud provider with no issues in delivery capacity—it operates over 100,000 self-owned servers, with H1 capex and leased assets totaling RMB 6.24 billion, and an adjusted EBITDA margin of 32.0%. The challenge lies in its loss-making track record (though losses are narrowing) and a ~71% debt-to-asset ratio, indicating high investment intensity.

Yue Gang Wan Smart Computing (01396.HK): Transformed from traditional businesses, its transformation is relatively comprehensive—operating across four lines: multi-source heterogeneous computing power integration, "Quantum Pai" computing power cloud services, domestic and overseas AIDC construction/operations, and AI vertical application investment incubation. Its model most closely resembles overseas Nebius's Neocloud path. AI business accounts for 93.1% of revenue, with over RMB 37 billion in intent orders (over RMB 20 billion billed and delivered), over 95% of new contracts being 3–5-year long-term agreements, stable utilization rates above 95%, and H1 net profit of RMB 281 million derived purely from operations.

Green New Energy (01783.HK): With a background in building construction and green energy, it has built a "circular economy + computing power" dual-mainstay business through acquisitions in recent years: Beyond its power battery recycling network, it acquired YV Cloud and its operator Shanghai Ucloud Computing (one of China's earliest NVIDIA Cloud Partners) in June 2024, entering computing power leasing. While its direction is clear, its computing power business remains in its infancy—its current PE (TTM) stands at ~249x, reflecting mainly market expectations.

Among these five, GDS Holdings, SUNeVision, Kingsoft Cloud, and Green New Energy are all included in the Stock Connect, while Yue Gang Wan Smart Computing is not—its daily turnover remains persistently low, reflecting liquidity differences.

Inclusion in the Stock Connect is a processual goal for companies. Market estimates suggest potential inclusion around 2027, subject to Hang Seng Indexes' review results and exchange announcements. While connectivity differences affect trading convenience, they do not alter fundamentals. However, they remind investors to judge non-Stock Connect stocks based on delivery and profitability.

[Conclusion]

The computing power industry's growth clearly cannot rely on a few leading firms alone.

As AI applications penetrate more industries, demand for computing power infrastructure, resource scheduling, cloud services, and other links continues to rise. Established players have years of accumulation, while new entrants can also carve out spaces with their resources and capabilities. Greater participation will enrich the industry's business models and expand its development potential.

However, computing power is not a business that yields quick results. From business layout and equipment deployment to order delivery, and then to forming stable revenues and profits, time is required. Especially for cross-border transformers, nurturing new businesses often involves sustained investments and repeated explorations, making it difficult to deliver perfect results shortly.

While the capital markets are justified in focusing on operational outcomes, excessive speculation divorced from fundamentals must also be guarded against. For firms actively transforming and seeking new growth spaces, their futures should not be prematurely dismissed due to short-term stock price volatility or earnings pressure.

After all, a thriving industry needs both mature firms that have already succeeded and new participants bold enough to enter the fray. For firms still transforming, beyond focusing on immediate results, the market should also grant them more time and patience.

Risk Warning: Data in this article is sourced from public market information and company disclosures (as of intraday trading on October 13, 2026) and is for informational sharing only, not investment advice. Individual stocks exhibit extreme volatility—please be aware of risks.

Disclaimer

The content concerning listed companies in this article represents the author's personal analysis and judgments based on publicly disclosed information (including but not limited to ad hoc announcements, periodic reports, and official interaction platforms) as required by law. The information or opinions herein do not constitute any investment or other commercial advice, and Market Cap Watch assumes no liability for any actions taken based on this article.

—END—