When Ant Lingbo seeks independent financing, the first thing that comes to mind is capital, but the driving force behind it is Ant's internal budget strategy.

According to BlueWhale Technology, the catalyst for this move is the group's substantial investment in AI, which has put a strain on its computing power budget.

Given Ant's actions this year, this development is not unexpected.

In 2025, Ant Group's investment in technological R&D soared to RMB 35.03 billion, reaching an all-time high in both scale and revenue proportion. That same year, Ant Group's net profit stood at approximately RMB 15.3 billion, a significant drop of about RMB 23 billion from the previous year, as new growth initiatives and tech investments took a toll on profits.

Nevertheless, these expenditures are warranted. Han Xinyi previously remarked that in the AI era, major companies are all making "saturated investments," and Ant has rejoined the fray.

The budget is primarily allocated to embodied intelligence, as Ant currently juggles numerous AI projects vying for resources.

Bailing Large Model, AI Payments, Ant Afu, Alipay AI Portal, Financial Large Model, and Embodied Intelligence all demand computing power, talent, and organizational focus. Embodied Intelligence, in particular, has even more stringent requirements. It involves not just training a model and launching a product but also real-world data acquisition, simulation system development, hardware adaptation, scenario pilot testing, and deployment teams.

Zhou Jun, Vice President of Ant Group, publicly noted that the computing power cost for a trillion-parameter model is exorbitant, with each 15-minute operation consuming the equivalent cost of a Tesla. While this statement grabs attention, its more practical implication is that internal budgets cannot be expanded indefinitely.

Lingbo, established in December 2024 with a registered capital of RMB 100 million, is wholly owned by Ant Group, with Zhu Xing serving as its legal representative. Over 90% of the company's technical talents hold master's or doctoral degrees. Within Ant, it operates on par with the Bailing Large Model team and the Alipay AI team, focusing on "physical intelligence."

If Lingbo remains within the group, it will compete with other AI teams for the same computing resources. Independent financing allows Lingbo to leverage external funds to buy time and must demonstrate its valuation, progress, and commercialization milestones to investors.

This approach is more conducive to securing additional resources during a critical window period.

Financial technology is adopting a similar strategy to secure more resources and is moving at a faster pace than embodied intelligence. In early 2024, Ant Group restructured its organization, establishing separate boards for Ant International, OceanBase, and Ant Digital Technologies, each operating independently and responsible for its own profits and losses. Subsequently, Ant International completed its Series A financing.

On July 29 of this year, multiple media outlets reported that both Ant Digital Technologies and OceanBase are undergoing new rounds of financing, with Ant Digital Technologies allegedly preparing for a Pre-IPO round.

Ant Digital Technologies did not issue a public response. However, sources familiar with the matter told Caijing that there are indeed internal plans for an IPO.

This development is not sudden. In September 2024, Zhao Wenbiao addressed questions about listing at a media exchange, stating that there was no timeline yet, but if "the entire business continues to grow," the time may not be far off. At that time, he had just served as CEO of Ant Digital Technologies for 170 days.

Ant Digital Technologies originated from the commercialization of technology within Ant.

It began as a blockchain lab in 2015 and later evolved into AntChain. In 2019, one year before Ant Group's attempted IPO, AntChain embarked on independent commercialization. Following the organizational upgrade in 2024, Ant Digital Technologies, Ant International, and OceanBase were positioned more independently.

Now, the question arises: does the capital market view Ant Digital Technologies as a financial technology company or an AI company?

According to Caijing, a tech sector investor bluntly stated that financial technology companies are valued at around 30 times PE, while AI companies command valuations exceeding 80 times. In his view, Ant Digital Technologies leans more towards the former, with limited room for stock price appreciation.

Ant Digital Technologies clearly does not wish to be pigeonholed as such.

In 2025, spurred by DeepSeek, financial institutions accelerated the adoption of large models. Ant Digital Technologies merged its AI and cloud businesses, shifting its focus to financial large models and intelligent agents.

The company now boasts three main businesses: financial large models and intelligent agents, financial risk control, and blockchain and privacy computing. Among these, financial risk control is growing steadily, while the blockchain business has seen a relatively subdued response due to regulatory policies. Financial large models and intelligent agents have emerged as key growth drivers.

According to an IDC report, the market size for financial large models, intelligent agent applications, and services in China reached RMB 2.436 billion in 2025, with Ant Digital Technologies leading the market with a 13.3% share. Ant Digital Technologies CTO Yan Ying stated that its intelligent agent services cover 100% of state-owned major banks and 60% of city commercial banks.

Finance has provided Ant Digital Technologies with clients, risk control experience, and compliance capabilities, but it has also imposed boundaries. This is a dilemma within Ant. Technology businesses emerge from the group, relying on the parent company for support in the early stages. However, when it comes time to go public, they must demonstrate that they are not mere appendages of the parent company.

Although Lingbo, which is smaller than Ant Digital Technologies, is still far from an IPO, it is capitalizing on a favorable financing window.

In the first half of 2026, total financing in the domestic embodied intelligence sector reached approximately RMB 43.8 billion, with companies focusing on embodied brains absorbing RMB 22.253 billion, accounting for 50.8%. As of mid-June 2026, there have been 88 financing events in China's embodied intelligence sector, with cumulative financing exceeding RMB 20 billion.

Backed by Ant Group's resources and the sector's high growth potential, Lingbo's financing should proceed smoothly, with its valuation "naturally at the unicorn level or higher."

Lingbo recently underwent a personnel change.

According to Red Interstellar Robotics, Luo Zhikun, the head of data at Ant Lingbo, has departed the company, and Chief Data Scientist Huang Yongtao will continue to lead the team. Luo Zhikun is a rare talent in the domestic embodied intelligence field with practical experience in closed-loop data systems. Huang Yongtao has a background in autonomous driving and was formerly a technical partner at Yuanrong Qixing, where he led the construction of autonomous driving data and training infrastructure.

The changes in the data team suggest that Lingbo is undertaking challenging tasks.

In January 2026, it continuously open-sourced four embodied large models. In early July, it released six more embodied intelligence large models within a week, covering vision, video, spatial perception, dexterous manipulation, world models, and world action models, all of which were open-sourced. It has also launched its first single-scenario service robot, Robbyant-R1, with its "robot smart pharmacy" solution selected as a "highlight of WAIC 2026" and collaborations with companies such as Leyou Robotics, Orbbec, and Jianzhi Robotics.

Lingbo has opted for a more resource-intensive approach, investing tens of thousands of hours in real-world data.

This strategy is more akin to that of autonomous driving: first establishing a data closed-loop, then iterating on models, and finally discussing large-scale scenario deployment.

Zhu Xing stated at the Zhiyuan Conference that embodied intelligence, similar to autonomous driving, is a long-cycle sector, and 2026 will witness small-scale commercial pilots in specific scenarios. Lingbo's Chief Scientist, Shen Yujun, also mentioned that there is no need to be overly concerned about technical routes, as the entire embodied brain is still in its infancy, with many issues to resolve.

In the early stages of embodied intelligence, model demos are easily visible to investors, but data systems are not. In 2025, many embodied companies were relatively cautious with data, either training on open-source datasets or conducting limited data collection for specific scenarios, then fine-tuning open-source models.

Internal projects at major companies have this advantage. However, startups must justify every additional dollar spent to investors in terms of potential orders; group projects can first establish infrastructure.

Once Lingbo seeks independent financing, this approach will be scrutinized by external investors. Questions will arise about whether tens of thousands of hours of real-world data can yield stable action capabilities and be replicated in more scenarios, and whether open-source models can foster an ecosystem.

Examining Ant Digital Technologies and Lingbo together, Ant is pursuing a strategy of allowing different AI businesses to be priced by the market in distinct ways.

Ant Digital Technologies is closer to revenue generation and IPO. It must demonstrate that its financial large models, intelligent agents, and risk control businesses can generate stable revenue and that it is not solely reliant on the Ant ecosystem for clients.

Lingbo is closer to the technological frontier. It must prove that embodied intelligence is not just a showcase project for major companies but a data and engineering route that can advance with external capital.

The two companies are at different stages but share the same internal premise: the group no longer covers all uncertainties with a single master budget.

In 2020, Ant attempted to include payments, financial technology, technical services, and innovative businesses in a single group IPO narrative. Now, this narrative has been dismantled. Ant International focuses on globalization, OceanBase on databases, Ant Digital Technologies on financial AI, and Ant Lingbo on physical AI.

Externally, this is perceived as a financing and IPO process. Internally, it is a resource prioritization exercise.

Ant is addressing questions about which businesses will continue to receive group resources, which will seek external capital, which will rush for an IPO, and which will endure in labs and pilot scenarios for a few more years. AI investments will persist, but the ledger is no longer kept solely within the group.