January 15, 2026
From Four Companies to Three Tech Pillars: How Stellar AGI Labs Builds a Full-Stack Map from Hong Kong
Four companies, three tech programs — how Stellar AGI Labs, starting from its first AI product in Hong Kong in 2023, has been building a full-stack technology map spanning AI, IT and Web3.
Should a tech company focus on doing one product well, or extend up and down the stack around a core problem?
Stellar AGI Labs chose the latter. Beginning with its first AI product launched in Hong Kong in 2023, the company has gradually moved into large models and APIs, enterprise software, AI compute and storage hardware, and Web3 trusted trading — establishing four companies across Hong Kong, Shenzhen and the United States.
This makes Stellar hard to label simply as a chatbot company, a software developer or a hardware brand. By its own positioning, it is a technology system spanning AI, IT and Web3, aiming to connect models, applications, enterprise services, underlying hardware and trusted-trading mechanisms.
This article is compiled from publicly available information from Stellar AGI Labs and AskAIs up to June 2026. User figures, technical capabilities and future plans are the company's publicly stated positions.
What Is Stellar AGI Labs?
Stellar AGI Labs is a brand and technology system shared by four operating companies, not a single product.
"Stellar" represents the Hong Kong headquarters, "AGI" represents a long-term vision for artificial intelligence, and "Labs" emphasizes continuous R&D and product experimentation. The company's core purpose is to raise productivity through AI and help Hong Kong become a hub of intelligent technology.
Bringing that macro vision back to the product level, Stellar mainly addresses two problems: AI and IT solve productivity, while Web3 solves trust in cross-border digital trading; models, APIs and hardware provide the underlying support for these products.
Launching the First AI Product from Hong Kong
In February 2023, Hong Kong Stellar Digital Technology Limited was established. On the 23rd of that month, the team launched its first AI product, chat.stellar.hk, offering conversation via GPT large models and intelligent image generation with Stable Diffusion.
Generative AI was just entering the mass market then. Rather than first challenging large foundation models, Stellar chose to turn existing model capabilities into a product ordinary users could use directly.
This choice let the team reach real users and learn what people wanted AI to do — and it built an entry point of demand for the later API, enterprise services and model R&D.
From Application into Models and API
In June 2023, Stellar released its first fine-tuned GPT large model and opened an API, beginning to build intelligence closer to real business by using open-source models and vertical-domain data for enterprises.
User-facing products solve "how people use AI"; the API solves "how other products gain AI". The two can form a loop: front-end applications discover needs, models and APIs turn capabilities into a platform, and more enterprise products plug in.
AskAIs LLM's public directions currently include self-developed models, vertical fine-tuning, industry-specific models, a unified API, private deployment and enterprise customization. The developer platform also offers multi-model interfaces, pay-as-you-go billing, load balancing, failover and web-augmented retrieval.
Building Mainland China Operations and Compliance
In 2024, Stellar established Shenzhen Xingwen Artificial Intelligence Co., Ltd. in Shenzhen, responsible for mainland China AI business, product operations and compliance. AI and API also got a separate mainland edition aligned with the local model environment, payment channels, content safety and regulatory requirements.
The AskAIs website currently displays large-model algorithm filing, ICP filing and value-added telecom business licence information. This shows the team incorporates regional compliance into the product architecture, rather than merely translating the same website into different languages.
From Software into AI Hardware
In 2025, Stellar began building an AI compute and storage hardware business. The company says external chip and supply restrictions had affected service stability, so it decided to reduce reliance on a single external supply chain.
The team established Shenzhen Xingwen Chip Technology Co., Ltd. in Shenzhen for manufacturing, while the Hong Kong R&D headquarters leads design — forming a "Hong Kong R&D, Shenzhen manufacturing" model.
The first public product is AskAIs Mini SSD. Officially positioned as M.2 NVMe 2280 storage for LLM training, offering 128GB to 4TB of capacity, its product page currently shows research is complete and it is preparing for launch.
That said, hardware and SaaS are entirely different industries. An SSD must ultimately prove its value through its controller, NAND, firmware, thermals, sustained read/write, endurance, mass-production yield and after-sales capability. An AI-training positioning also needs complete specifications, third-party testing and real-workload support.
Going Global and the M&N Model Program
In 2026, Stellar established MiniCode LLC in the United States, responsible for international-market operations and global compliance. With this, the company forms a three-region structure across Hong Kong, Shenzhen and the United States.
In May that year, Stellar launched the M&N Large Model R&D Program in Hong Kong, focusing on a general-purpose large model, a multimodal large model and a Hong Kong-style language model.
The Hong Kong-style language model is the most locally distinctive direction. Global models can handle Chinese, but may not truly understand written Cantonese, mixed English-Chinese expression, Hong Kong industry terminology and local cultural context. With high-quality corpora, evaluation standards and enterprise scenarios, this capability could serve Hong Kong's customer service, finance, retail, media and public services.
How Do the Four Companies Divide Roles?
Hong Kong Stellar Digital Technology Limited is the headquarters, responsible for Hong Kong and Asia operations, product, brand and some R&D; Shenzhen Xingwen Artificial Intelligence Co., Ltd. handles mainland China AI products and compliance; Shenzhen Xingwen Chip Technology Co., Ltd. handles hardware manufacturing; and MiniCode LLC handles the international market.
This structure combines Hong Kong's international connectivity, Shenzhen's engineering and supply chain, and the US's global market.
But the four companies also need unified management of intellectual property, brand, technical standards, cross-border data, customer contracts and financial relationships. The more products and the larger the scale, the more important formal group governance becomes.
Pillar One: Artificial Intelligence
AI is Stellar's most mature business, including AskAIs AI for individuals and teams, AskAIs LLM and the API for enterprises and developers, the M&N model R&D program, and AI compute and storage hardware.
According to figures published by the company, AskAIs surpassed 1 million total annual users in 2025, with cumulative API calls exceeding 10 billion. User numbers reflect product reach, while API call volume reflects the high-frequency machine requests the interface bears.
These two figures mainly represent AskAIs and should not be taken to mean every Stellar product line has reached the same maturity. The next phase deserves more attention to monthly active users, paid conversion, enterprise customers, service availability and task completion rate.
Pillar Two: Enterprise IT
Stellar's IT business includes enterprise SaaS and software product licensing, covering intelligent customer service, team collaboration, business management, official websites and mobile apps.
Though less eye-catching than foundation models, this business may be a crucial link in commercializing technology. Enterprises do not automatically complete digitalization just because a model is advanced; they still need accounts, permissions, back-office, support, processes and system integration.
The IT business also lets Stellar get close to enterprise customers, feeding real needs back into the AskAIs API, intelligent customer service and vertical models.
Pillar Three: Web3 Trusted Trading
Stellar's Web3 plan centers on WeTruste. The product is still in development, planned to serve cross-border virtual goods and service trading such as software source code, white-label systems and software development.
By its public design, WeTruste will combine third-party escrow, blockchain records, smart contracts, seller verification, dispute arbitration and transaction traceability, protecting both parties until delivery is complete.
But blockchain itself cannot automatically establish trust. The platform must still solve identity review, escrow qualification, payment compliance, refunds, intellectual property and human arbitration. WeTruste's ultimate competitiveness will not come merely from "going on-chain", but from complete trading rules and risk management.
Can the Three Businesses Create Synergy?
AI, IT and Web3 can be understood as a complete chain from production and delivery to trading: AI raises content and work efficiency, IT deploys capabilities into enterprise systems, and Web3 attempts to solve payment and trust for cross-border digital services.
Ideally, an enterprise could use Stellar's software to build a website and customer service, add AI via the AskAIs API, and then sell cross-border digital services through WeTruste.
But true synergy must be quantifiable: does the same customer buy multiple services? Does the AskAIs API lower SaaS R&D costs? Do IT customers become WeTruste merchants? Does hardware genuinely improve model service cost and stability?
Only when these relationships are productized do the three businesses form a platform, rather than just three categories on a website.
Why Is Hong Kong Stellar's Headquarters?
Hong Kong has a bilingual Chinese-English business environment and connects mainland China, Asia and international markets. Professional services, cross-border contracts, intellectual property and international payment systems make it suitable as a coordination window for tech companies.
Stellar has also not treated Hong Kong merely as a place of registration and brand. The M&N model program, the Hong Kong-style language model, the hardware R&D headquarters and AskAIs' Asia operations all show the company wants to keep product and R&D capabilities in Hong Kong.
If it can in future work with Hong Kong universities, enterprises, media and public institutions to build local corpora and evaluation systems, Hong Kong will be not just an office but a source of Stellar's differentiated technology.
Opportunities and Risks of a Full-Stack Path
Stellar's advantage is that it can discover real needs through AskAIs, amplify capabilities via API and enterprise software, and then raise self-reliance with models and hardware. Hong Kong, Shenzhen and the US also offer complementary conditions.
The risk is that the scope of business may exceed the organization's capacity. AI applications, foundation models, SaaS, software development, SSDs, chips and a trading platform each need specialist teams, capital, compliance and long-term investment.
"Self-developed" also needs a clear explanation of technical boundaries. Which parts Stellar designs, which use partner solutions, and how core IP and performance are verified will all affect professional markets' trust in the brand.
Conclusion: Full-Stack Is Not Doing Everything
Looking back at 2023 to 2026, Stellar's growth has not been simply adding products.
It first turned large models into user products in Hong Kong, then opened capabilities as an API; it then built independent operations and compliance for mainland China, extended into AI storage hardware, and took on the international market with its US company.
The four companies represent different regions and capabilities, while the three tech programs point respectively to efficiency, digitalization and trust. Whether they can become a true technology group depends on whether the different businesses share customers, technology and supply chains, and whether clear governance can turn synergy into results.
Full-stack does not mean doing everything; it means that every step forward makes the core product more reliable, more controllable and closer to real problems.
If Stellar can keep starting from real products and customer needs, Hong Kong will be not just its headquarters, but possibly the starting point of a locally distinctive, globally oriented technology and product system.