On the evening of August 25, full-stack intelligent product and service provider iSoftStone released its 2026 semi-annual report. In the first half of the year, the company achieved revenue of RMB 18.607 billion, up 17.91% year-on-year; Q2 revenue reached RMB 10.490 billion, up 19.61% year-on-year, with net profit attributable to shareholders of RMB 85 million, up 54.71% year-on-year.

More important than revenue and profit figures is the full delivery of iSoftStone's AI strategy. In the first half of the year, iSoftStone's AI-related business revenue reached RMB 11.435 billion, accounting for 61.46%, up 46.61% year-on-year, setting new historical highs in both scale and growth rate. Revenue from computing products and smart electronics reached RMB 9.479 billion, exceeding 50% for the first time, up 40.32% year-on-year, becoming the primary driver of the company's performance growth.

All signs indicate that iSoftStone's 'full-stack intelligence' strategy is entering a harvest phase, and the AI industry dividend is just beginning.

Hardware Revenue Exceeds 50% for the First Time, 'Hardcore Implementation' of AI Strategy

In the first half of 2026, revenue from computing products and smart electronics reached RMB 9.479 billion, exceeding 50% for the first time, up over 40% year-on-year. Against the backdrop of acquiring Tsinghua Tongfang's computer business, iSoftStone's hardware revenue share has continuously improve ( continuously improve translates as 'continuously increased' in context, kept as is for specificity) in recent years, rising from 41.26% in 2024 and 45.05% in 2025 to breaking 50% in the first half of this year.

This reflects the essence of 'bidirectional empowerment': on one hand, external M&A has indeed elevated iSoftStone's revenue scale to a new level, achieving 'recreating an iSoftStone'; on the other hand, iSoftStone's mature supporting software development capabilities, deep customer accumulation, and brand recognition have successfully revitalized Tongfang's computer business, achieving significant business expansion.

It should be noted that AI's rapid penetration has created new conditions for hardware expansion. The 'general-purpose computing + intelligent computing' matrix built by iSoftStone Huafang covers full-scenario computing needs. The Super series intelligent computing servers support multiple domestic accelerator chips such as Ascend 950PR, providing a computing base for intelligent computing centers; the Super series general-purpose servers are equipped with Kunpeng 920 processors, becoming the main models for government cloud and enterprise private cloud construction; high-performance, large-scale air-cooled inference super-node servers have been deployed in some provinces and cities; terminal business covers six domestic CPU architectures. The essence of these products is to provide computing infrastructure for AI—hardware growth is a direct reflection of surging AI demand.

The same is true on the consumer side. Mechanical Revolution accounted for 18% of the domestic gaming laptop market share in the first half of the year; the Xingyao 14 ultrabook and Yaoshi 16 Ultra gaming laptop repeatedly topped e-commerce sales charts, with the core driver of growth being the rapid implementation of AIPC products, with multiple models passing national terminal intelligence L3 certification. Ultimately, every 'blockbuster' sold is an AI terminal.

Computing products exceeding 50% is not a victory for traditional hardware but a 'proof of implementation' of iSoftStone's AI strategy in the physical world.

AI-Related Business Revenue Reaches RMB 11.4 Billion in Half a Year, Share Hits New High

In the first half of the year, iSoftStone's AI-related business revenue reached RMB 11.435 billion, up 46.61% year-on-year, setting new historical highs in both absolute scale and growth rate. Baima Analytics previously stated that iSoftStone's 'AI strategy has entered a harvest phase,' and the semi-annual report confirms this—the harvest is coming faster than expected.

More noteworthy is the rising 'quality' of AI business. Leveraging its AI Factory full-stack capabilities, iSoftStone has built six major product and service capabilities: intelligent computing services, computing products, Token resale, FDE, smart terminals, and Token self-use, closing the AI cooperation loop between model vendors and internet companies.

In June this year, iSoftStone's self-built 'Beijing No.1 Token Factory' went live in Beijing, achieving a breakthrough in Token factory operations from 0 to 1; signed intelligent computing service agreements with leading large model vendors to provide Token inference services; the Guangdong-Hong Kong-Macao Greater Bay Area public computing service platform and Token factory officially launched; the Ruidong Intelligent Agent Cloud Platform completed commercial validation of Token resale from 0 to 1. iSoftStone has become a bona fide 'AI Factory,' providing comprehensive AI hardware and software services to customers.

Strategic Investments in SiliconFlow and Tanwei Xinlian Extend AI Layout Deeply

Of course, iSoftStone is not content with the present but is further deepening its AI layout. In the first half of this year, the company made consecutive investments in SiliconFlow and Tanwei Xinlian, enhancing 'full-stack intelligence' industrial synergy.

Public information shows that SiliconFlow is an AI inference service company positioned as an 'open and independent Token supply platform,' using self-developed technology to uniformly schedule various AI chips and provide customers with efficient, low-cost AI computing (Token) services; Tanwei Xinlian originates from Tsinghua University's Brain-Inspired Computing Research Center and is the first domestic company to fully stack-compete with NVIDIA's 'NVLink+NVSwitch' architecture. Its core product is a fully self-developed Scale-up super-node communication interconnection solution aimed at solving multi-chip communication bottlenecks in AI large model training.

These two companies—one bolstering 'efficiency infrastructure for model inference' and the other enhancing 'domestic layout of computing chips'—have enabled iSoftStone to take another step upstream in the 'computing power—model—application' chain.

Baima believes that evaluating iSoftStone's AI layout requires looking beyond revenue figures in financial reports to its industrial depth. From building its own Token factory to strategically investing in model infrastructure and chip companies, to co-establishing the Artificial Intelligence Systems Joint Research Center with Tsinghua University's Computer Science Department and launching the 'National Model, National Chip' open-source zone with the Ministry of Industry and Information Technology's Fifth Electronics Institute—iSoftStone is upgrading from an 'AI service provider' to a 'deep participant in the AI industrial chain.' This positioning offers short-term investment returns and long-term Bind ( Bind translates as 'binding' in context, kept as is for specificity) to the upstream lifeblood of the entire AI industry.

Coupled with an innovation ecosystem of 'industry-academia-research + venture capital + customer labs,' iSoftStone's AI moat is deepening.

Three Major Businesses Synergize, AI Flywheel Begins to Spin

Now, stepping back to fundamentals, the three major businesses—software, hardware, and intelligent computing services—have formed a clear 'AI flywheel.'

Software and digital technology services serve as the starting point, with FDE frontline delivery + three-platform support + model computing power supply, landing benchmark projects in over ten industries including finance, manufacturing, agriculture, and steel: in agriculture, signing over RMB 10 million in contracts with clients like Wens Group and Shuangbao in the first half of the year; in finance, applying large models and intelligent agents for penetrating regulation and intelligent risk control; in manufacturing, transportation, and energy, forming replicable solutions such as smart ports and industrial AI quality inspection.

Computing products and smart electronics act as the lever, with hardware growth meeting AI demand; digital energy and intelligent computing services serve as the foundation, with green power direct connection, computing centers, and Token factories Connect the entire chain ( Connect the entire chain translates as 'connecting the entire chain' in context, kept as is for specificity) for computing-power synergy. These three elements interlock, and iSoftStone's AI flywheel has indeed started spinning.

Industry standing is solidifying simultaneously. iSoftStone was successfully selected for the 2026 Fortune China Top 50 Tech Companies, ranked 412th on the Fortune China 500, retained its top position in the CCID Consulting China IT Services Market and IT Consulting Domestic Vendors list, was selected for the IDC 2026 China AI 50, ranked TOP5 in the 'IDC China AI Professional Services Market Insights 2025H2' (TOP4 in platform and application solution services, TOP5 in infrastructure integration services), and ranked third in notebook shipments and fourth in desktop shipments in the government and education sectors in Q1 2026 for iSoftStone Computers.

Internationally, Mechanical Revolution achieved breakthroughs in the U.S. and Philippines, secured large-scale educational smart terminal procurement projects in South Asia, Southeast Asia, and Africa, and launched AI benchmark projects in Malaysia, Bangladesh, and other regions. The dual drivers of 'AI digital technology services + smart terminal products' are clearly driving overseas growth.

AI is telling different stories across industrial chain links. For iSoftStone, AI is not a concept but is delivering tangible results—AI-related business revenue surpassing 60% and continuing rapid growth; simultaneously, computing products as AI's hardware carrier are scaling up, with strategic investments upstream strengthening influence. iSoftStone's 'AI tree' is continuously blooming and bearing fruit.

Article content and opinions are for reference only and do not constitute investment advice. Investment involves risks; decisions should be made cautiously.