51World Deep Dive: A Listed Digital Twin Company Where 51Sim Becomes the First Growth Engine While 51Earth Nearly Vanishes, and Two Irreconcilable Financing Naming Conventions Must Both Be Kept

Executive Summary

Beijing 51World Digital Twin Technology Co., Ltd. is a Beijing company that listed on the Hong Kong Stock Exchange on 30 December 2025 under stock code 6651. It is one of the very few pure-play digital twin companies in the Chinese mainland that reached a public market, and the interesting analytical question is no longer whether the business exists. It is whether the revenue mix that the listing prospectus described is still the revenue mix that the company reports in mid-2026, because in the space of one reporting period the composition of the business changed more than the headline growth rate suggests.

The headline growth is real and strong. Revenue for the first half of 2026 reached CNY 124 million, up 129.8% year on year, with gross margin rising from 41.1% to 44.8%. Research and development spending in the same period was CNY 54.3 million, up 80.3%. The loss narrowed: the net loss for the period was CNY 70.05 million and the loss attributable to shareholders was CNY 67.57 million, 25.5% narrower than a year earlier. For a company that has never reported an annual profit, that combination is the strongest operating print in its history.

Underneath the growth, the business lines have diverged to the point where the strategy has changed shape. 51Aes, the digital twin platform and the founding product, contributed CNY 72.33 million, up 65.2%, and remains the largest single line at roughly 58% of revenue. 51Sim, the synthetic data and simulation business built around its SimOne product, contributed CNY 51.24 million, up 545.2%, and now accounts for 41.4% of revenue against 14.8% in the same period a year earlier. 51Earth, the digital earth business launched in 2022 and positioned as the company's strategic flagship for a planetary-scale model, contributed CNY 103,000, against CNY 2.094 million in the same period a year earlier, a decline of 95.08%. A business line that was supposed to carry the long-term narrative is now contributing less than 0.1% of revenue, and the company did not allocate any of the IPO proceeds to it in the first half of 2026.

The financing history cannot be presented in one timeline because two incompatible naming conventions circulate for the same company. The company's own materials describe an A round in July 2015, a B round in December 2017, a C round in March 2020, a C-plus round in November 2021 and a pre-IPO round in August 2024. A widely circulated financial-press account of the same company describes the identical sequence as A, A-plus, B, B-one, C, D, E and F rounds with completely different amounts, and the two schemes share no label at all beyond the letters A, B and C. This report keeps both and does not reconcile them into a single table.

The listed financials are unusually well documented for a company of this type. Revenue for 2022, 2023 and 2024 was CNY 170 million, CNY 256 million and CNY 287 million respectively, a compound annual growth rate of 30.02%. Net loss over the same three years was CNY 190 million, CNY 87 million and CNY 79 million, and the adjusted net loss was CNY 132 million, CNY 68 million, CNY 43 million and CNY 94 million for 2022, 2023, 2024 and the first half of 2025 respectively. These are prospectus-grade numbers, and they are the strongest evidentiary asset this report has.

The market has already priced a great deal of that evidence. The shares opened at HK$35 on listing day, 14.75% above the HK$30.5 offer price, and market capitalisation briefly exceeded HK$14 billion. By 18 August 2026 the figure quoted in market coverage was around HK$27.35 billion, and another observation puts it at HK$32.17 billion with a price-to-book ratio of 39.85 times. Against a 2026 half-year revenue of CNY 124 million, those valuations are extremely sensitive to the forecast, which is why the sell-side number matters so much.

The single most consequential external opinion is Guosheng Securities, which initiated coverage with a buy rating on 20 October 2026 and a target price of HK$51.3. The valuation method is a sum-of-the-parts built on 12 times 2027 estimated price-to-sales. The revenue forecasts are CNY 750 million, CNY 1.53 billion and CNY 2.62 billion for 2026, 2027 and 2028, implying growth of 117%, 103% and 71%, with non-GAAP net profit of negative CNY 110 million, positive CNY 10 million and positive CNY 250 million. That target sits below the August market quotation, which is worth noticing: the published target does not support the price the market was paying in the summer.

Three structural features deserve attention. The first is the founder incentive structure: the chairman and chief executive's salary for 2025 to 2030 is capped at around HK$510,000 per year, and full vesting of his equity incentive requires the market capitalisation to reach HK$100 billion, roughly three and a half times the August 2026 level. The second is index inclusion: the Hang Seng Composite Index announced inclusion in February 2026, implemented after the close on 6 March and effective from 9 March, with China International Capital Corporation noting the possibility of a subsequent Stock Connect addition. The third is customer breadth, with management claiming more than one thousand large and medium enterprises across nineteen countries and regions, and Frost and Sullivan ranking the company first in China by 2024 digital twin solution revenue.

The overall judgement is that this is a real business with real, verifiable growth and a documented financial history, attached to a strategy narrative that has quietly been overtaken by its own product mix. The technology claims remain largely self-reported, the international expansion is asserted rather than evidenced, and the valuation has run well ahead of the forecast.

Company Overview

The listed entity is Beijing 51World Digital Twin Technology Co., Ltd., a joint stock company incorporated in the People's Republic of China and listed in Hong Kong under stock code 6651 with a listing date of 30 December 2025. The business originates in a corporate reorganisation: the corporate registry records incorporation on 16 February 2015, while the group's own materials describe the company's founding as February 2015, and a predecessor entity named Dangjia Mobile Internet Green Property was registered on 22 October 2014 under the real estate group Modern Land. The real estate ancestry is not a side note but a material fact about the customer base, and it is treated here as part of the analysis rather than as a curiosity.

The founder is the dominant individual and the disclosure around him is unusually specific. Li Yi holds a cost engineering degree from Chongqing University dating from 2008 and an MBA from Tsinghua University awarded in June 2023. He became chief executive in 2015, chairman in November 2016 and general manager in April 2018. Between 2008 and 2013 he worked at First Property and First Mcode, both real estate companies. That combination, a construction-cost background with a residential real estate operating history, is unusual for a simulation company and it partly explains the company's early client concentration in property and urban planning, which is also the origin of the 51Earth digital earth line.

The product architecture has three named pillars plus a new embodied layer. 51Aes, the digital twin platform, was launched in 2015 and remains the company's foundation. 51Sim, the synthetic data and simulation business whose core product is SimOne, was launched in 2017 and has become the growth engine. 51Earth, the digital earth business, was launched in 2022. On 19 August 2026 the company released AperData and AperOne, an embodied intelligence layer organised around three application modes named Testing, Training and Operations. The company positions these four as a progression from a digital twin of built assets, through synthetic data generation, through a digital earth, to embodied agents that act on the physical world.

Headcount and internal scale are disclosed only in fragments. The company reports more than one thousand large and medium enterprise customers across nineteen countries and regions, and describes 51Sim coverage of 55% of the global top twenty passenger vehicle original equipment manufacturers and 60% of the Chinese top twenty, together with six national-level Chinese testing and evaluation institutions as of the end of 2025. No consolidated employee headcount, no R&D headcount and no revenue-per-employee figure appears in the public record for the listed period.

Shareholding and governance after listing are only partly disclosed. The founder holds a substantial position and exercises options over 7.6476 million shares, about 1.8% of the share capital, exercised on 13 April 2026. The full post-listing shareholder register, free float percentage and lock-up expiry schedule are not available in the public sources consulted.

Common Misreadings: Four Errors That Change the Analysis

The company did not start from a clean engineering background. Tracing the business back to a 2014 real estate subsidiary of Modern Land is correct and material, but it does not mean the simulation business was a repurposing exercise. 51Aes was launched in 2015 and the digital twin positioning was there from the start; the property heritage explains the client base and the 51Earth lineage rather than undermining the technical direction.

51Earth is not a growth driver. Any summary describing the company as a digital twin company with three balanced pillars is out of date. In the first half of 2026 the digital earth line contributed CNY 103,000, down 95.08% year on year and below 0.1% of revenue, and it received zero of the CNY 109 million of IPO proceeds deployed during the period. The company's own capital allocation contradicts the narrative that the digital earth is its strategic centre.

51Sim is not a synthetic data side business. It contributed 41.4% of first-half 2026 revenue, up 545.2% year on year, and has overtaken the valuation of the narrative. Any competitive comparison that treats the company as a digital twin vendor missing the simulation wave has the direction of the risk exactly backwards.

The company does not have a single financing timeline. The rounds described as A, B, C, C-plus and pre-IPO by the company and the rounds described as A, A-plus, B, B-one, C, D, E and F by a financial press outlet refer to the same underlying transactions with different labels. Neither list can be used to count rounds, and any statement of the form "51World has raised X in total" must first specify which convention it follows.

Technology

The technical claim of 51World is that a single spatial data foundation can serve simulation, twin and embodied application layers, and the four product lines are the concrete expressions of that claim.

51Aes is the spatial data and twin platform. It ingests BIM, CAD and geospatial data, maintains a live model of physical assets and exposes twin state to downstream applications. Its value proposition in the listing documents is enterprise workflow integration, that is, embedding twin state into design, construction, operations and maintenance processes rather than delivering a visualisation product.

51Sim is the synthetic data and simulation business, with SimOne as its core product. Synthetic data generation for perception training is the technical core: producing images, point clouds and video-like data for systems that cannot safely collect enough real-world samples. The 2026 revenue acceleration of 545.2% sits directly on top of this product, and the reported customer coverage of major original equipment manufacturers and national testing institutions is consistent with that reading. The company also positions 51Sim as a general simulation platform serving industrial and automotive users for design exploration and validation, which is the segment that overlaps most directly with the model-first cohort of physics AI vendors.

51Earth is the digital earth business, that is, a planetary-scale spatial model intended to support global positioning, navigation, timing and remote sensing applications. Technically it is the most ambitious of the three and commercially the least developed. Its revenue collapse in the first half of 2026 is the single most important commercial datum in this report.

AperData and AperOne, released on 19 August 2026, are the embodied layer. AperData is the data and environment side, and AperOne is the model and agent side, organised around Testing, Training and Operations. The logical connection is that a twin and a synthetic data generator already supply the environments and demonstrations an embodied policy needs, so the same customer that buys 51Sim for perception training is the natural buyer of an embodied training stack. This is the most coherent product argument the company has, and it is also the newest and least evidenced.

The evidence base for accuracy and performance is thin. The public record contains customer outcome figures and market-share rankings but essentially no published benchmark: no mesh convergence study, no error band against a reference solver, no reproducibility package. The Frost and Sullivan first-place ranking for 2024 digital twin solution revenue in China is a commissioned market-research conclusion whose underlying methodology and revenue definition are not disclosed, and it measures revenue in a market the company itself helped define. Reported platform metrics such as simulation speed-ups are customer-reported and carry no test matrix.

The relationship with physical simulation is integration rather than new solver development. The company does not disclose an in-house finite element, finite volume or lattice Boltzmann solver. The technical contribution is best described as the spatial data layer, the scenario generation and synthetic data pipeline, and the application integration, layered on top of third-party solvers and on the neural operator and surrogate model literature that the wider field has produced.

Product and Commercial Model

The commercial model has three revenue lines with very different economics, and their divergence is the story of the past eighteen months.

51Aes remains the anchor. CNY 72.33 million in the first half of 2026, up 65.2%, roughly 58% of revenue. The client profile is large enterprises and public-sector owners of physical assets, sold through a project-and-platform mix that carries implementation cost. The 2026 first-half report shows the company allocating HK$53.3 million of IPO proceeds to 51Aes research and development, by far the largest allocation.

51Sim has become the second pillar and the fastest. CNY 51.24 million, up 545.2%, 41.4% of revenue against 14.8% a year earlier. The relevant capital allocation is HK$10.7 million, which is proportionally modest and suggests the growth is being driven by platform reuse rather than by heavy investment. The revenue composition here is worth pricing: synthetic data for perception training is a genuine market with genuine budgets, and it is adjacent to, but not the same as, high-fidelity engineering simulation.

51Earth is a rounding error. CNY 103,000, down 95.08%, zero IPO allocation in the period. A strategic business line with no capital and no revenue is effectively a research option, whatever the prospectus said.

The capital structure after the listing is the clearest statement of intent. The offering was 23.975 million shares at HK$30.5, raising HK$731 million gross and HK$650 million net, with the over-allotment option taking the gross figure up to as much as HK$841 million. Of the money deployed in the first half of 2026, CNY 109 million went into research and development across the three lines, with 51Earth receiving nothing.

Market position is asserted more than demonstrated. Management claims more than one thousand large and medium enterprise customers across nineteen countries and regions, and the Frost and Sullivan ranking as China's largest digital twin solution provider by 2024 revenue. Neither claim comes with a customer count that distinguishes paid customers from pilots, nor with a revenue split by geography outside China, and the international footprint in particular is asserted through a country count rather than through named overseas references.

The one hard external anchor is the Guosheng Securities initiation of 20 October 2026. Buy rating, target price HK$51.3, sum-of-the-parts on 12 times 2027 estimated price-to-sales, revenue forecasts of CNY 750 million, CNY 1.53 billion and CNY 2.62 billion for 2026 to 2028 with non-GAAP net profit of negative CNY 110 million, positive CNY 10 million and positive CNY 250 million. The target implies the market price observed in August 2026 was above the analyst's own fair value, which is a rare and informative configuration in a newly listed name.

Competitive Landscape

The company operates across three different competitive sets at once, and they should not be merged.

Against digital twin platform vendors, the comparison is feature and workflow completeness in asset-heavy industries. The 51Aes position, deep integration with design and construction processes and with public-sector asset owners, is a real differentiator against pure visualisation tools, and the Frost and Sullivan revenue ranking is the strongest single competitive artefact the company holds. The weakness is that this segment is project-shaped, which limits operating leverage and makes the gross margin trajectory dependent on delivery mix rather than on software.

Against synthetic data and perception-training vendors, the 51Sim position is materially stronger than it was two years ago, and this is where the 545.2% growth comes from. The competitors here are not primarily the CAE vendors but the specialist synthetic data and simulation platforms, plus the in-house capability of the original equipment manufacturers themselves. The structural risk is that the largest customers are also the most capable of building this internally.

Against physics AI and simulation foundation model vendors, the company's position is unusual and worth stating precisely. It is a customer-side, workflow-side and data-side participant in physical simulation rather than a supplier of simulation engines or learned operators. That means the model-first cohort does not directly displace 51Sim's twin and synthetic data business; it can, however, absorb the design-exploration layer that 51Sim also sells, and it can make the underlying physics calls that 51Sim's own products surface. This is a genuine adjacency risk that the 2026 results have so far outpaced.

On the international stage, the relevant comparisons are with Siemens and Dassault Systèmes digital twin businesses and with NVIDIA Omniverse. Those platforms arrive with existing CAD and simulation franchises, so the contest for enterprise wallet share is asymmetric in their favour, while 51World's counterweight is delivery cost and local engineering service in the Chinese market.

In humanoid robotics, the AperOne embodied layer places the company in a very crowded and very well-funded field, competing for the same budgets that Physical Intelligence, Skild AI, Figure and 1X are pursuing, and it does so from a position of no disclosed robotics hardware, no disclosed manipulation success rates and no published embodied benchmark.

Financing

The company listed on the Hong Kong Stock Exchange on 30 December 2025. The offering was 23.975 million shares at HK$30.5 each, gross proceeds of HK$731 million, rising to as much as HK$841 million with the over-allotment option, and net proceeds of HK$650 million. This is the one financing event for which the company, the exchange and the underwriters all give the same numbers.

The pre-listing rounds exist under two naming conventions that cannot be merged. The company's own account is: A round in July 2015 with USD 5 million from Guanghe Venture Capital, Tujia, Modern Land and WeiMu; B round in December 2017 of CNY 210 million from SenseTime, Modern Technology, Green Pine, StarVC and Guanghe; C round in March 2020 from First Asset, Oriental Pearl, China Merchants Securities Investment and Yunjiu; C-plus round in November 2021 with Moore Threads; and a pre-IPO round in August 2024 of CNY 200 million from Nanning Huanxin Capital. The financial press account of the same company is: A round of USD 3 million at a post-money valuation of USD 13 million, A-plus of USD 8 million at USD 40 million, B of USD 20 million at USD 140 million, B-one of USD 1 million at USD 151 million, C of CNY 183 million at a post-money CNY 1.783 billion, D of CNY 120 million at CNY 3.62 billion, E of CNY 96.42 million at CNY 4.096 billion and F of CNY 200 million at CNY 4.4 billion. Both lists describe one company and one sequence of transactions. This report reports them separately and does not add them together.

What can be stated without ambiguity is the capital raised at listing, the identity of the pre-listing investors, and the fact that the company has never reported an annual profit.

The founder's incentive structure is the most unusual financing-adjacent feature in the file. Li Yi's total salary for 2025 to 2030 is contractually capped at around HK$510,000 per year, and full vesting of his equity incentive requires the market capitalisation of the company to reach HK$100 billion. The August 2026 market capitalisation of roughly HK$27.35 billion therefore leaves a substantial distance to run, and the option exercise of 7.6476 million shares on 13 April 2026 is a partial step.

DISCREPANCIES

1. Financing round naming. Convention A, the company's own materials: A (July 2015), B (December 2017), C (March 2020), C-plus (November 2021), pre-IPO (August 2024). Convention B, financial press: A, A-plus, B, B-one, C, D, E, F. Both are official or media sourced, and they share only the letters A, B and C. Ruling: both conventions are reported separately and are not merged into one timeline. A is the company's own disclosure and is preferred for the transaction sequence; B carries the only disclosed post-money valuation figures and is preferred for valuation history. Neither count of rounds is used as the number of rounds the company has completed.

2. Series A size. Convention A: USD 5 million, July 2015 (OFFICIAL company materials). Convention B: USD 3 million at a post-money valuation of USD 13 million (MEDIA). Ruling: the two are not averaged. The difference between the two figures, USD 3 million against USD 5 million, is large enough that one of the two excludes some tranche or the other includes one, and no source explains which. This report quotes the company figure as the company figure and the press figure as the press figure.

3. Series C-plus against Series D. Convention A: C-plus of November 2021 with Moore Threads participating (OFFICIAL). Convention B: Series D of CNY 120 million at a CNY 3.62 billion post-money (MEDIA). Ruling: treated as the same transaction with incompatible labels, and stated as such. Whether Moore Threads invested at this point in the C-plus round or in a separate round is not resolvable from public sources.

4. Gross proceeds at listing. Convention A: HK$731 million (OFFICIAL). Convention B: up to HK$841 million (OFFICIAL, over-allotment). Ruling: both are correct under different definitions. HK$731 million is the base offering; HK$841 million is the maximum including the over-allotment option; net proceeds of HK$650 million are the figure actually available for use. The three must never be quoted interchangeably as the size of the raise.

5. Market capitalisation. Convention A: above HK$14 billion briefly on listing day (MEDIA). Convention B: around HK$27.35 billion on 18 August 2026 (MEDIA). Convention C: HK$32.17 billion at another observation, with a price-to-book ratio of 39.85 times (MEDIA). Ruling: these are different observation dates on the same series, not conflicting accounts of one moment. No single figure is presented as the market capitalisation, and each quotation is always attached to its date.

6. 51Earth strategic status. Convention A: the digital earth is a strategic pillar and a core part of the long-term model (company narrative and prospectus framing). Convention B: the line contributed CNY 103,000 in the first half of 2026, down 95.08%, and received zero of the CNY 109 million of IPO proceeds deployed (OFFICIAL financial disclosure). Ruling: the financial disclosure is controlling on present commercial weight, and the strategic framing is retained as stated intent. Both are reported, and no attempt is made to reconcile a strategic pillar with 0.07% of revenue.

7. Revenue coverage of the top twenty vehicle manufacturers. Convention A: 55% of the global top twenty passenger vehicle original equipment manufacturers (company statement). Convention B: 60% of the Chinese top twenty (company statement). Ruling: both are retained as separate global and domestic figures. The two are close enough to be mistaken for a conflict, but they describe different populations and no source reconciles them.

8. Hang Seng Composite Index effective date. Convention A: announcement on 13 February 2026 (OFFICIAL index notice). Convention B: implemented after the close on 6 March and effective from 9 March (OFFICIAL index notice). Convention C: a commentary that the inclusion may lead to a Stock Connect addition (broker research). Ruling: the announcement, implementation and effective dates are three distinct events and are reported as three. A potential Stock Connect inclusion is a forecast and is not written as an accomplished fact.

9. Loss figures for the first half of 2026. Convention A: net loss for the period of CNY 70.05 million (OFFICIAL). Convention B: loss attributable to shareholders of CNY 67.57 million (OFFICIAL). Ruling: both, with the reason given. The difference is attributable to non-controlling interests and is not a contradiction.

10. 2025 first-half revenue base. Convention A: CNY 54 million for the first half of 2025 (OFFICIAL). Convention B: the implied prior-year base derived from the reported 129.8% growth. Ruling: A is used, and the implied figure is not computed separately. The two are consistent to rounding and the prospectus figure is the disclosed one.

Risks and Due Diligence Gaps

The digital earth narrative is exposed. A line that fell 95.08% year on year and received no IPO allocation is either an option with no buyer or a business whose previous disclosure overstated its traction. Either reading is a material reputational and disclosure-quality signal about the other lines.

No solver of its own means no control over the physics. The company's technical contribution is the data, twin and synthetic data layer. If model-first vendors make physics calls cheaper and better, the value of the layer above them is contested, and the company has no engine with which to defend the stack.

Synthetic data is the segment where the largest customers are also the best equipped. 51Sim's growth depends on large original equipment manufacturers and national testing institutions continuing to prefer an external vendor for perception training data over building it in house. This is the most likely source of revenue concentration and price pressure, and no customer concentration disclosure exists.

Overseas expansion is asserted, not evidenced. A count of nineteen countries and regions is not a reference list. Without named overseas customers, revenue split or renewal data, the international claim cannot be weighed.

Losses are narrowing but the company has never been profitable. Cumulative losses across 2022 to 2024 alone were CNY 356 million on the reported basis, and the first half of 2026 still lost CNY 70.05 million. The non-GAAP break-even forecast arrives only in 2027 under the most bullish published model, and that model assumes revenue more than doubles in 2027.

Valuation sensitivity is extreme. A price-to-book ratio of 39.85 times at one observation, and a published target price below the market price of the previous month, together define a name in which the expectation gap is wide. Any deceleration in the 51Sim growth rate should be expected to move the price substantially.

The founder's HK$100 billion vesting condition is a concentration risk and a governance question. It aligns incentives with a very long and very aggressive outcome, but it also means a large part of the equity story depends on one individual's continued tenure and on a valuation level the company has never approached.

The two financing conventions still coexist three years after listing. A listed company whose own account of its pre-IPO rounds cannot be reconciled with the widely published version is a disclosure-quality red flag, and it means no third party can state the company's total pre-IPO capital raised with confidence.

Technology claims have no published verification. There is no accuracy benchmark, no convergence study, no error band against reference solutions and no third-party reproduction. For a company whose differentiation rests on simulation fidelity, this is the largest gap in the file.

Overall Assessment

• Technology 6.5. A genuine spatial data foundation with a working digital twin platform, a synthetic data and simulation business that has proven product-market fit with major vehicle manufacturers and national testing bodies, a coherent new embodied layer built on the existing data assets, and unusually complete listed financial disclosure. Deducted points: no in-house solver, no published accuracy benchmark or convergence study, no reproducible performance test matrix, the digital earth line in effective retreat, and the newest product layer being the least evidenced.
• Commercialisation 6.0. A CNY 287 million revenue base in 2024 growing to 129.8% year on year in the first half of 2026, gross margin improving from 41.1% to 44.8%, a mix shift toward the higher-value simulation and synthetic data line, and a first half of 2026 that crossed CNY 124 million. Deducted points: no customer concentration disclosure, overseas footprint asserted through a country count, project-shaped legacy revenue limiting operating leverage, no disclosed recurring versus project revenue split, and no audited multi-year order book.
• Financing 6.5. A completed Hong Kong listing raising HK$731 million gross and HK$650 million net, a pre-IPO investor list that includes SenseTime, Moore Threads and several state-linked and market investors, disclosed use of proceeds, and a founder incentive structure unusually aligned with long-term value. Deducted points: no cumulative pre-IPO total can be stated with confidence because two naming conventions coexist, no public disclosure of pre-IPO valuation from the company itself, losses across every year of the disclosed history, and no dividend or buyback capacity.
• Product maturity 7.0. Four product layers with a clear internal logic, the newest one released only four months before this report, a functioning commercial simulation product inside the fastest growing line, an embodied intelligence layer organised around Testing, Training and Operations, and a mature enterprise twin product with ten years of delivery history. Deducted points: the flagship earth layer commercially dormant, the embodied layer with no disclosed benchmark or partner hardware, and no published platform-level performance or accuracy figures.
• Market position 6.0. A top-ranked position in Chinese digital twin by commissioned market research, index inclusion in the Hang Seng Composite with a possible Stock Connect path, a listed currency that makes the company visible to international investors, and a genuine adjacency to the synthetic data market for embodied AI training. Deducted points: competing against Siemens, Dassault and NVIDIA with a fraction of their resources, no named overseas reference, valuation far ahead of the current revenue base, and a published target price below the observed market price.

References

1. 51World prospectus and listing documents, HKEX stock code 6651, including incorporation on 16 February 2015, the offering of 23.975 million shares at HK$30.5, gross proceeds of HK$731 million and net proceeds of HK$650 million
2. 51World 2026 interim results announcement, including revenue of CNY 124 million up 129.8%, gross margin of 44.8%, net loss of CNY 70.05 million, loss attributable to shareholders of CNY 67.57 million and R&D of CNY 54.3 million up 80.3%
3. 51World segment disclosure for the first half of 2026, including 51Aes revenue of CNY 72.33 million up 65.2%, 51Sim revenue of CNY 51.24 million up 545.2% at 41.4% of revenue, and 51Earth revenue of CNY 103 thousand down 95.08%
4. 51World use of proceeds disclosure for the first half of 2026, including CNY 109 million deployed, HK$53.3 million to 51Aes, HK$10.7 million to 51Sim and zero to 51Earth
5. 51World 2024 annual results, including revenue of CNY 170 million, CNY 256 million and CNY 287 million for 2022 to 2024 with a 30.02% CAGR, net loss of CNY 190 million, CNY 87 million and CNY 79 million, and adjusted net loss of CNY 132 million, CNY 68 million, CNY 43 million and CNY 94 million
6. 51World company and product materials on 51Aes, 51Sim, SimOne and 51Earth, including the 2015, 2017 and 2022 launch dates and the 19 August 2026 release of AperData and AperOne with Testing, Training and Operations modes
7. International Finance News and Moore Threads listing coverage, containing the alternative A, A-plus, B, B-one, C, D, E and F round convention with post-money valuation figures, dated prior to the December 2025 listing
8. Frost and Sullivan 2024 China digital twin solution revenue ranking as reported in 51World materials, with the methodology and revenue definition not disclosed
9. Guosheng Securities initiation of coverage, 20 October 2026, including the buy rating, HK$51.3 target price, sum-of-the-parts on 12 times 2027 estimated price-to-sales, and revenue and non-GAAP profit forecasts for 2026 to 2028
10. Hang Seng Indexes announcement of 13 February 2026 on Hang Seng Composite Index inclusion, implemented after the close on 6 March and effective from 9 March, with the Stock Connect observation from China International Capital Corporation
11. Market quotation records for listing day and August 2026, including the HK$35 opening price, the market capitalisation observations of around HK$14 billion, around HK$27.35 billion and HK$32.17 billion, and the price-to-book ratio of 39.85 times
12. 51World founder and remuneration disclosures, including the 2025 to 2030 salary cap of around HK$510,000, the HK$100 billion market capitalisation vesting condition, and the exercise of 7.6476 million shares on 13 April 2026

This report is compiled from public sources. It is for reference only and does not constitute investment advice. Data is current as of October 23, 2026; please refer to official company announcements and filings.