Feldspar Capital: With "70% Pre-Investment Services," uncover the most certain "Alpha" in the embodied intelligence track

Ask AI · How does LongStone Capital’s 70% pre-investment service identify high-quality founders with embodied intelligence?

The “spring” of financing in the embodied intelligence track has already arrived.

Since March, a wave of embodied intelligence companies have successively completed financing: first, on March 4th, SynaTech, focusing on logistics scenario embodied robots, announced the completion of over 200 million yuan in strategic financing; next, on March 11th, Litian Intelligent, specializing in photovoltaic energy intelligent robots, disclosed over 100 million yuan in Pre-B round financing.

Meanwhile, at the end of March, Shihang Intelligent and Differential Intelligence also announced new rounds of financing. Among them, Shihang Intelligent completed strategic financing of several hundred million yuan in A+ and A++ rounds; Differential Intelligence completed a few hundred million yuan in A1 round financing.

In just one month, many vertical companies in the embodied intelligence sector have “officially announced” their funding, igniting market enthusiasm and sending a strong signal: the commercialization of embodied intelligence in specific scenarios is accelerating.

By dissecting these independent financing events and carefully tracing the capital flow, we find a common name behind them: LongStone Capital. As the early or key investor in these four companies, LongStone Capital plays a crucial “catalyst” role.

Recently, Wang Gongbin, founding partner of LongStone Capital, said in an exclusive interview with PitchBook that the competition in the second half of embodied intelligence will not be limited to the parameter battles of general large models, but will focus on the efficiency and overall cost of “robots” replacing “people” in specific scenarios, as well as a billion-dollar market with imagination.

Against this backdrop, LongStone Capital is laying out the entire AI industry chain around the “foundation and ultimate form of embodied intelligence,” starting from the ultimate energy, computing, and carriers needed by AI, and expanding into fields such as controlled nuclear fusion, quantum computing, AI infrastructure, embodied intelligence, AI hardware, and AI agents.

Multiple recent investments verify the “vertical scenario value”

The collective breakout of these four companies this spring of 2026 is essentially a focused validation of the “closed-loop commercialization of vertical scenarios.”

In LongStone Capital’s investment map, Shihang Intelligent, Differential Intelligence, SynaTech, and Litian Intelligent represent four major vertical segments: underwater, low-altitude, logistics, and photovoltaics. Their rapid growth confirms LongStone Capital’s previous precise judgment: the value of vertical applications lies in achieving stronger, more cost-effective, and safer replacements in specific scenarios, breaking industry pain points in non-standard scenes through technological breakthroughs, and building unique barriers to commercialization.

Take SynaTech as an example. This company has completed five rounds of financing so far, with top institutions like China Merchants Venture Capital, Alibaba, and Dachen Financial Intelligence investing. LongStone Capital first invested in December 2025 and has continued to increase its stake.

SynaTech’s appeal to industry capital and professional institutions mainly stems from its precise tackling of the non-standard “last 20 meters” problem in logistics loading and unloading, and its early construction of a complete “R&D–Customer Expansion–Profit Model” commercialization loop. Its logistics-specific large model, built through nearly 10,000 hours of real-world training and data accumulation, paired with self-developed arms, mobile platforms, and flexible end-effectors, allows industrial robots to perfectly adapt to unstructured logistics environments with cluttered cargo and dynamic operations. Relying on industry know-how accumulated in logistics, it has formed a data, technology, and industry experience triple barrier that is difficult to replicate, enabling all-weather human replacement in logistics loading and unloading, significantly reducing operational costs, and securing orders from multiple state-owned enterprise giants.

Shihang Intelligent is similar. Although this company focuses on oceanic applications—an area rarely explored—since its founding in 2023, it has completed seven rounds of financing. LongStone Capital participated in its Series A in September 2025 and increased its investment in November of the same year. Its business has successfully entered supply chains of China Merchants Ship and China Nuclear Group.

Faced with extreme environments like ocean currents, high salinity, high pressure, and corrosion, Shihang Intelligent believes that, compared to the previous purely data-driven approach, a hybrid robust architecture combining physical models, real-time perception, and adaptive learning is more suitable for environments where communication is impossible, ensuring stable, accurate, and reliable operations, solving pain points in ship maintenance and marine infrastructure.

Shihang Intelligent sees their selection by state-owned enterprises as a choice of practical, implementable, and stable long-term services. “Our approach of ‘full-stack self-research + extreme scenario validation + creating value for clients’ has earned recognition from state-owned enterprise clients.”

Litian Intelligent and Differential Intelligence also follow the logic of “reliable landing in non-standard scenarios.” Litian Intelligent focuses on the full lifecycle of photovoltaic energy, with application scenarios in the Middle East where surface temperatures reach 50°C, in Australian mining areas plagued by sandstorms, or in South Africa’s complex terrain. Relying on real operational data from multiple GW-scale projects worldwide to iterate algorithms and hardware, Litian Intelligent leverages its “technology R&D + massive on-site data + engineering capability” integrated advantage to enable long-term stable operation of its robots under extreme conditions like high temperature, sandstorms, and rugged terrain. Its photovoltaic installation robots, which can replace 3-5 workers with a single machine, greatly shorten the construction cycle of solar power stations, transforming the mode of building new energy power plants.

Differential Intelligence targets complex scenarios such as mining, forestry, power, and emergency rescue. Confronted with extreme environments and the core needs of fully autonomous operations across scenes, it has developed a embodied intelligence general technology platform, using a “general platform + lightweight scene adaptation” model to enable rapid cross-scenario technology reuse and large-scale deployment, directly addressing high-risk, hard-to-reach operational pain points. After its initial investment in September 2025, LongStone Capital increased its stake again in January this year.

Overall, these four companies invested by LongStone Capital precisely target the core pain points of “people can’t do it, high labor costs, low efficiency.” According to Wang Gongbin, the core of vertical embodied intelligence is to achieve a robot cost ≤ 50% of human labor cost and robot efficiency ≥ 10 times human efficiency, enabling true business value from “R&D” to “market application.” Under this logic, these four companies have already crossed the critical threshold from “laboratory to scenario deployment.”

In 70% pre-investment services, find the most certain “Alpha”

If precise positioning in vertical scenarios is LongStone Capital’s approach in embodied intelligence, then its unique “70% pre-investment service” model is the core secret to continuously discovering high-quality targets and helping companies grow rapidly. Wang Gongbin hopes to find the most certain Alpha (α) in the industry’s upward Beta (β) through LongStone’s exclusive methodology.

This methodology is not arbitrary but the result of LongStone Capital’s continuous validation and iteration since its all-in commitment to hard tech in 2017: from “Smartphones + AI” to “Smart Cars + AI,” and now to “Embodied Intelligence.” At each stage, LongStone insists on “deep industry chain research as the foundation, using pre-investment services to identify founders with business sensitivity, and funneling core resources into the most certain vertical scenarios.” This practical, battle-hardened approach keeps LongStone’s judgment on the implementation pace of hard tech industry landing precise.

So, how does the “70% pre-investment service” work? In LongStone’s view, investment is never just capital injection but a deep industry symbiosis. The core of this model is to focus 70% of effort on pre-investment—not traditional due diligence or risk control, but investing with a “service mindset,” proactively helping potential portfolio companies connect with customer resources, adapt to industry needs, and build a multi-party win-win industrial network.

Wang Gongbin states that the essence of investment is investing in people. The “Alpha” in LongStone’s view is founders with high business sensitivity; the 70% pre-investment service is about identifying this high-quality Alpha gene. During this phase, the team does not stay in the office obsessing over technical details but actively uses industry resources, repeatedly communicates with founders, tech teams, and customers to refine products, verify market fit, and observe whether founders can seize industry clues and turn resource connections into actual business productivity.

For example, when deploying SynaTech, LongStone pre-connected top logistics clients for real-world testing; when engaging with Shihang Intelligent, it also coordinated resources to connect upstream and downstream supply chains. This “pre-empowerment” service philosophy allows LongStone to accurately identify high-quality founders early and avoid the gap from lab to commercialization.

To ensure industry resources are precisely empowered to the most valuable companies, LongStone has built a multi-layer funnel screening mechanism: from hundreds of initial contacts, it first filters based on industry logic, scenario pain points, and landing standards to select about ten promising targets, then allocates core service resources to them; after deep empowerment and repeated validation, investments are made, ensuring resource input is targeted from the source.

Meanwhile, LongStone’s unique “Hundred-Person Industry Resource + Investment Research Driven” dual-wheel model supports this methodology. On one side, the “Hundred-Person” gathers founders and executives from listed companies like Huawei, OPPO, BYD, Zhaosheng Micro, and Sunlord Electronics—these industry LPs are not only investors but also “industry radar,” deeply involved in supply chains, providing high-quality information before project financing, enabling LongStone to deploy high-quality targets at low valuations.

On the other side, independent investment research is the foundation of LongStone’s judgment. Wang Gongbin explains that each year, the team sets research topics, exploring from market deepening and industry top-down perspectives, visiting companies, and mapping industry chains to produce in-depth reports; all colleagues participate in multiple research groups to improve project judgment through cross perspectives.

Regarding independent judgment, Wang Gongbin uses a vivid analogy: “A person’s ability circle depends on the average level of the six people they communicate with most frequently and closely. LongStone hopes that among these six, five are from industry and one from investment, so that the ability circle is closest to industry—making judgments based on frontline industry information, supplemented by independent research to avoid excessive industry noise.”

This long-term tracking and repeated validation have enabled LongStone to form a three-dimensional understanding of potential investments before they are made. As Wang Gongbin quotes, “Conscious targeting, unconscious stimulation.” When a company needs financing, LongStone can make quick decisions and act efficiently.

So far, this precise judgment and deep empowerment have yielded tangible results: the performance and valuation of invested companies have both grown. Companies like Shihang Intelligent and SynaTech have doubled or even multiplied their revenues, with valuations continuously rising, becoming benchmarks in the vertical embodied intelligence field.

Covering the entire AI industry chain from infrastructure to terminal applications

In fact, since 2026, the total financing amount in the embodied intelligence track within two months has exceeded 20 billion yuan, compared to 2025’s total of 73.54B yuan, indicating a very hot capital market.

In response to this wave, Wang Gongbin believes that behind it lies solid industry logic, but also some bubbles. The real opportunities are hidden in precise positioning across the entire industry chain, deep cultivation of vertical scenarios, and dual bets on industry resources and high-quality founders.

Based on this, Wang Gongbin makes a clear judgment: if last year was the year of soaring valuations for general embodied intelligence, then 2026 will be the year of valuation increases for vertical embodied intelligence. “Compared to last year’s hype in general tracks, this year vertical scenario companies are more likely to produce ‘billion-dollar clubs.’”

But bubbles should also be guarded against. Currently, the track shows a pattern of “valuation growth > performance growth.” In the long run, as application scenarios expand, performance growth will eventually restore valuation levels. The biggest market bubble lies in some companies failing to find real application scenarios, lacking customer stickiness, and relying solely on hype.

In Wang Gongbin’s view, the embodied intelligence track has moved past the initial exploration stage. Industry reshuffling is inevitable. “In the process, leading companies may strengthen their position through expanding application scenarios and increasing customer stickiness, forming a pattern of strong getting stronger; those that haven’t locked in vertical applications and lag behind the leaders will face pressure.”

Faced with opportunities and challenges, LongStone Capital is prepared: from infrastructure to terminal applications, deploying across the entire industry chain. Especially in quantum computing, a frontier field, PitchBook has learned that LongStone has made early moves. Wang Gongbin revealed that LongStone has invested in two companies in this area, one of which is a quantum measurement and control chip company that earned over 60 million yuan last year and is expected to surpass 100 million this year.

When discussing the investment logic of quantum computing, Wang Gongbin said that its value lies in solving complex problems that traditional computing cannot handle, providing core support for AI efficiency leaps; as the “ultimate computing” track, quantum computing has both high Beta (β) and high Alpha (α) attributes. Several companies have achieved commercialization revenue domestically and abroad, with successful listings in the US stock market, making exit paths clear.

Therefore, LongStone plans to allocate 20% of its funds to frontier fields like controlled nuclear fusion and quantum computing, with the remaining focus on AI infrastructure and vertical scenarios in embodied intelligence.

Reflecting on years of investment practice, Wang Gongbin states frankly: “Investing must start with the end in mind,” meaning thinking about exit paths at the time of investment. Among LongStone’s first fund’s 24 projects, 12 have gone public, with 9 listed on the STAR Market. The 50% IPO rate is driven by two core principles: first, thorough industry chain mapping during pre-investment to ensure each project is among the top three in its industry; second, seizing timing opportunities based on industry cycles and policy windows.

Of course, the effective operation of this methodology depends on continuous internal review. Monthly review meetings involve re-examining invested projects, analyzing deviations from expectations, and clarifying strategies for follow-up investments, exits, or maintenance, dynamically adjusting to optimize asset allocation.

Looking ahead, Wang Gongbin is full of confidence. He believes that the demand for human replacement is rigid, and the integration of technology and scenarios is inevitable. LongStone will continue to deepen its focus on vertical embodied intelligence, promote core component domestic substitution, and accompany high-quality companies through cycles, creating long-term value in the wave of AI and physical world integration.