51WORLD Long-Term Value Research Report

From a long-term owner's perspective, I score this business's "understandability" at 3/5. The technical concepts are not simple, but the commercial logic is not complicated: sell platforms, sell solutions, do integration, do simulation, do data. What truly pulls my score down is not that I "can't understand the technology" but that I still cannot see clearly which economic model it ultimately wants to become: a low-margin, low-predictability project company, or a high-margin, scalable platform company? If the stock market closed for five years and I had to hold the whole company at the current price, I would not be willing to.

Industry, Competition, and Moat

The industry itself is attractive. Citing Frost & Sullivan data, the prospectus states that China's digital-twin design and operations-optimization solutions market is expected to grow from RMB 7.5 billion in 2024 to RMB 30 billion in 2029, a 32.0% compound growth rate; China's digital-twin AI training and validation solutions market is expected to grow from RMB 4.4 billion in 2024 to RMB 18.6 billion in 2029, a 33.5% compound growth rate. The demand drivers behind this space are also clear: digital upgrading of cities and infrastructure, autonomous-driving simulation testing, synthetic-data needs, robotics testing, and embodied-intelligence training. Looking only at the demand side, this is an industry with room to grow on a ten-year horizon.

But fast industry growth does not mean the company's moat is already solid. By 2024 revenue, 51WORLD held only a 2.4% share of China's overall digital-twin solutions market, a 3.2% share in the digital-twin design and operations-optimization segment, and a 1.1% share in the digital-twin AI training and validation solutions market. These figures tell me two things: first, although the company is one of the leaders under the broad market definition, it is far from a "monopoly"; second, this industry as a whole remains highly fragmented, and the profit pool is not firmly locked up by a few winners.

What truly deserves a careful look is the narrower 51Sim. In a voluntary announcement in March 2026, the company disclosed that, according to Frost & Sullivan's "China Physical AI Simulation and Data Platform Research Report," 51Sim ranked first in China's end-to-end advanced autonomous-driving simulation and data platform market with a 53.5% share; at the same time, the company said that in 2025 the number of its partnerships with China's top ten automakers rose from 4 to 8, reaching 80% coverage, and that it had partnered with 6 national-level authoritative testing institutions, reaching 100% coverage. If this set of figures keeps being validated by real future revenue and cash flow, 51Sim's local barriers would be clearly stronger than the company's overall business.

But it is essential to keep definitional discipline. The fact is: in a more narrowly defined market, 51Sim may already have built a very strong position. The inference is: this could bring stronger data accumulation, customer embedding, and ecosystem-niche advantages. What cannot be concluded directly is: that the company already has a broad moat across "the entire digital-twin / physical-AI industry." The reason is simple — in the company's 2025 group revenue, 51Sim made up only 16%, while 51Aes still accounted for close to 79%; and the company's latest statements still report under a "single operating segment," so investors cannot directly see each business line's margin, renewal rate, customer retention, or return on capital. The hottest narrative is not the same as the most important financial reality.

If I break the moat down item by item, my judgment is as follows. Brand advantage: some brand recognition within the digital-twin and autonomous-driving simulation circles, but not yet enough to form broad pricing power. Cost advantage: no sufficient evidence; on the contrary, 2025 gross margin fell sharply. Scale advantage: scale is still limited under the broad market definition. Network effects: may be starting to appear in 51Sim's data, scenarios, and partner ecosystem, but not yet proven at the group cash-flow level. Switching costs: if a customer has already integrated the platform into its R&D workflow or city/park operating systems, switching is not easy, but much of the revenue today is still project-type, so switching costs are not as firm as in PLM/ERP. Channel advantage: the network of system integrators and automakers/testing institutions is a strength, but it also brings the problem of margins being eaten by the channel. Patents, licenses, regulatory barriers: public materials are enough to show the company has technical accumulation, but not enough for me to confirm this intellectual property has turned into a solid economic monopoly. Data advantage: for 51Sim I assign a "has potential but unproven" verdict.

On balance, I give the industry 4/5 and the moat 2.5/5. It looks more like an early leader in a good industry than a great business that has already matured and can be safely valued as a "compounding machine." Whether it can raise prices in an inflationary environment depends, in my view, on whether the standardization and platformization transition succeeds; in an economic downturn it would most likely survive, because it holds ample net cash, but it may not stay profitable, since both public projects and automaker R&D budgets are pro-cyclical.

Management and Capital Allocation

On the management profile, 51WORLD is not a story of "hiring professional managers to dress it up for listing." Founder and CEO Li Yi has served as chief executive since 2015, chairman since 2016, and general manager since 2018; he has a real-estate operations and asset-management background and earned an MBA from Tsinghua University in 2023. Among the management team, several executive directors responsible for 51Aes, strategy execution, brand, and industry research have also mostly served inside the company for a long time. This management combination is not necessarily perfect, but it at least has the continuity of sustained entrepreneurship and continuous business execution.

On alignment of equity interests, the founder is indeed "tied to shareholders." The prospectus shows that, assuming the over-allotment option and pre-IPO options are not further issued, Li Yi will control about 23.3% of the company's voting rights after listing and hold about 25.0% beneficial interest. The annual report further discloses that as of December 31, 2025, Li Yi held a long position in about 24.98% of the company's interests in aggregate, including shares held directly and through controlled corporations; and on April 13, 2026, 7,647,600 H shares had been issued upon the exercise of his pre-IPO share options. From the angle of "whether the founder is in the same boat as shareholders," this is a positive.

On the direction of capital allocation, I give a verdict of "broadly rational, but not disciplined enough." The rational part is that the company has clearly designated the planned use of net IPO proceeds: 80% into R&D, allocated respectively to 51Aes, 51Sim, and 51Earth; 10% for marketing in China and overseas; and 10% for working capital and general corporate purposes. For a company still in the technology- and product-validation phase, directing most of its new-share financing into R&D rather than acquisitions and non-core assets is reasonable. As of year-end 2025 the company had also paid no dividends, conducted no buybacks, made no large acquisitions, and issued no convertibles — at least it has made no obvious moves to "beautify per-share metrics."

What I am truly unhappy about is that the price of dilution is too low and the constraints are too soft. In April 2026, the company granted a total of 940,200 RSUs to 303 directors and employees, at a grant price of RMB 0, and with no performance targets; the company explained this was mainly to incentivize and retain people. More importantly, the monthly report shows that as of April 30, 2026, 30.5905 million options remained unexercised under the pre-IPO share-option plan, equal to about 7.4% of total shares outstanding at the time. The 7,647,600 shares exercised that same month carried an exercise price of only HK$5.1, while the closing price on the RSU grant date was already HK$50.9. This means that shareholders bear not only operating-execution risk but also a fairly certain per-share dilution risk.

On governance transparency, I lean neutral-to-positive. The company's 2025 report did not dodge issues such as the gross-margin decline, rising R&D expenses, expanded receivables impairment, and negative operating cash flow, and KPMG issued a standard unqualified audit opinion on the 2025 financials. The annual report also discloses that the pre-purchase contract for the Beijing office building signed in 2024 was terminated in February 2026, with RMB 36 million in deposit refunded. This at least shows management did not quickly lock cash into heavy assets unrelated to the core business soon after listing. On the other hand, the company still reports under a single operating segment, making it hard for outside investors to judge how much real profit 51Sim actually creates, which lowers my assessment of its "full transparency."

On balance, I score management and capital allocation at 3/5. Looking only at direction, I would say "passing"; adding the item "whether it is friendly enough to growth in per-share intrinsic value," I can only say "not yet reassuring."

Financial Quality and Owner Earnings

First, the four-year history that can already be verified by public, audited data. The company's revenue from 2022 to 2025 was RMB 170 million, RMB 256 million, RMB 287 million, and RMB 348 million respectively, a roughly 26.9% three-year compound growth rate; but over the same period, gross margin fell from 65.0% to 54.2%, then to 51.1%, and dropped further to 30.0% in 2025. Net income was negative in all four years, with the 2025 loss widening to RMB 186 million. This shows that although the company is growing, this growth has not brought economies of scale; on the contrary, the 2025 growth came with worse unit economics.

The table below puts together the financial metrics I consider most critical. A note is in order: 2022–2025 is the main audited/disclosed history I can verify; the public basis for earlier years is insufficient and cannot be hard-coded. In addition, because the 2022–2023 capital-expenditure detail is incomplete in the materials I can verify this time, for free cash flow I only present the 2024–2025 figures that can be computed directly from the annual report's cash flow statement.

Note: 2022–2025 revenue, gross profit, losses, balance-sheet items, and net assets are from the "financial summary" in the 2025 annual report; 2025 operating cash flow, capital expenditure, cash, and borrowings are from the three primary statements in the 2025 annual report; 2022–2023 operating cash flow and period-end cash are from the cash flow summary in the prospectus.

For a long-term investor, the most jarring thing is not the "loss" but that no reassuring divergence has appeared between the loss and the cash flow. Many growth stocks can be explained as "accounting profit looks temporarily ugly, but operating cash flow is fine"; 51WORLD is currently not in that situation. The company's operating cash flow stayed negative from 2022 to 2025, and in 2025 the net operating cash outflow was RMB 91.75 million; after deducting RMB 1.9 million of payments for purchased materials/equipment and intangible assets, free cash flow was about -RMB 93.65 million. In other words, from the "business owner's" perspective, this company is still consuming cash today, not distributing it. It is certainly not short of cash — at year-end 2025 it held RMB 797 million in cash and cash equivalents, plus RMB 118 million in other financial assets, and after deducting RMB 286 million in bank loans it is still clearly in a net-cash position — but this safety comes mainly from listing financing, not from operating self-generation.

On profit quality, I have not found evidence in public materials that points directly to financial fraud, but receivables and impairment must be tracked closely. In 2025 the company's total trade receivables rose from RMB 214 million to RMB 320 million, and the related impairment provision rose from RMB 34.33 million to RMB 69.54 million; the impairment loss on trade and other receivables and contract assets recognized during the year reached RMB 41.32 million, nearly double that of 2024. In the annual report management also frankly admits that the main reason for the rise in impairment was the increase in receivables in 2025. For a project-type B2B company, long collection periods are not abnormal; but for a long-term shareholder, if revenue cannot be cashed smoothly, profit is not real profit.

Next, return on capital. Because net assets were negative in 2022, the early ROE basis is unstable; even so, the 2024 and 2025 ROE and ROA were essentially significantly negative. The net-debt/EBITDA item is also largely meaningless, because the company is in a net-cash position, yet EBITDA is still weak and even insufficient to support traditional multiples. Interest coverage is likewise of no reference value, because operating profit is negative. In other words, this company has not yet crossed the threshold of "deserving to be valued with a mature-business framework."

My "Owner Earnings" definition is more conservative. I use operating cash flow minus maintenance capital expenditure as the baseline; because software R&D is already expensed in the income statement, I no longer capitalize it separately; at the same time, I also do not add back equity incentives as distributable cash, because that would ignore real dilution. On this basis, 2025 Owner Earnings was about -RMB 93.65 million. Even if one insists on adding back all of the roughly RMB 39.07 million of 2025 equity incentives, the loss only narrows to about -RMB 54.58 million, still not turning positive, and this calculation is not honest about per-share value. The conclusion is direct: the current valuation is not "a few times owner earnings" — owner earnings are still negative.

My final judgment on financial quality is: the balance sheet is safe in the short term, but the income statement and cash flow statement do not yet have the quality a long-term owner wants. In an economic downturn the company would most likely survive; but "surviving" and "compounding for shareholders" are two different things.

Intrinsic Value, Margin of Safety, and Opportunity Cost

Here I first separate "fact, assumption, inference, and opinion." Fact: as of May 22, 2026, the share price was about HK$98.55; as of April 30, 2026, total shares outstanding were about 414 million; 2025 revenue was RMB 348 million, year-end net cash was about RMB 628 million, and Owner Earnings remain negative. Assumption: for valuation conversion, this report uses HK$1 ≈ RMB 0.92; the Owner Earnings discount model uses a 10-year explicit period. Inference: only when 51Sim truly converts its narrow-niche advantage into high-margin, recurring, reusable platform revenue can the company turn from a "project-type growth stock" into a "cash-flow-type compounding stock." Opinion: until this is proven by the financial statements over the next few years, the current price carries no margin of safety.

First, Method One: the Owner Earnings discount method. Because current Owner Earnings are negative, I can only do a scenario valuation in which they "turn from negative to positive over the next ten years," and this kind of valuation is inherently very sensitive to assumptions. My three scenarios are as follows: Conservative scenario: 10-year revenue CAGR 18%, year-10 free-cash-flow margin 8%, discount rate 12%, terminal growth rate 3%; Base scenario: 10-year revenue CAGR 28%, year-10 free-cash-flow margin 15%, discount rate 12%, terminal growth rate 3%; Optimistic scenario: 10-year revenue CAGR 38%, year-10 free-cash-flow margin 20%, discount rate 11%, terminal growth rate 3.5%. Under these three sets of assumptions, the per-share intrinsic value I derive is roughly HK$3–4, HK$9–10, and HK$28–30. The meaning of this result is not "the company is worth only this much," but rather: even granting a fairly aggressive growth curve, today's price has already bought a very distant success in advance.

Next, Method Two: relative valuation. Because 51WORLD's own net income, EBITDA, and free cash flow are all negative, PE, EV/EBITDA, and P/FCF have very weak reference value at this stage; I therefore put P/S first and use PB as a secondary aid. Roughly computing from the latest public market cap and 2025 revenue, 51WORLD currently trades at about 108x P/S and 51x PB. For comparison: GEOVIS, on 2025 revenue of RMB 2.677 billion and a market cap of about RMB 43.3 billion, trades at roughly 16x P/S; Suochen Technology, on 2025 revenue of RMB 466 million and a market cap of about RMB 9.58 billion, at roughly 21x P/S; Beijing DataTang, on 2025 revenue of RMB 377 million and a market cap of about RMB 8.6 billion, at roughly 23x P/S. Even though these comparables still carry very high PEs and thin profits, they are at least profitable or close to breakeven; 51WORLD's current valuation is clearly more expensive.

If we also bring in the market's hottest comparable AI / spatial-intelligence new listings for reference, the conclusion does not change. Manycore Tech's updated prospectus disclosed 2025 revenue of about RMB 820 million; its market cap after the first-day close was about HK$31.6 billion. Even at this already heated price, its P/S is only in the low thirties, still clearly below 51WORLD's current sales multiple of more than a hundred times. In other words, 51WORLD is not "expensive because peers are all expensive"; it is "even within an expensive peer group, it looks more expensive still."

Finally, Method Three: asset or liquidation value. 51WORLD is not a heavy-asset business. The prospectus once disclosed the company owns no real property; the annual report further shows that the pre-purchase contract for the Beijing office building signed in 2024 was terminated in February 2026, with RMB 36 million in deposit refunded. This means the support an asset approach provides comes mainly from the recoverable value of cash, financial assets, and receivables. Estimating on a conservative basis: cash and cash equivalents of RMB 797 million, other financial assets of RMB 118 million, restricted cash of RMB 23 million, recovering receivables and inventory at a discount and deducting all liabilities, liquidation value is roughly RMB 500 million to 700 million, equal to only about HK$1.4 to HK$1.9 per share; even looking directly at book net assets of RMB 732 million, book value per share is only around HK$1.9. The asset approach clearly cannot provide any support for the current price.

Based on the three methods, I give the following intrinsic-value ranges:

Under this framework, the current share price of HK$98.55 is a significant premium to my fair value range; even against the optimistic value range, it is still a large premium. For an investor who is balanced-to-conservative with a holding period of more than ten years, the margin-of-safety conclusion is very clear: insufficient, and almost nonexistent.

Opportunity cost does not support buying it now either. On May 22, 2026, China's 10-year government bond yield was about 1.75%. By my 10-year scenario model, buying at the current price gives annualized returns of roughly -21%, -8%, and +3% under the conservative/base/optimistic scenarios. In other words, only a very optimistic execution assumption barely reaches an unremarkable positive return; that is neither clearly superior to government bonds nor clearly superior to a diversified broad-based index. If I could only hold 5 assets, it does not qualify for the portfolio.

Risks, Checklist, and Final Conclusion

The biggest risk in this investment is not "share-price volatility" but permanent loss of capital. The most central risks fall into several categories. First, competition and substitution risk: market share under the broad digital-twin definition is not high, the industry is fragmented, and competition will persist long-term; if 51Sim's lead stays only within the narrow definition without converting into group-level profit and cash flow, the market will re-rate it quickly. Second, business-model risk: revenue is still mainly project-based, recurring revenue is not yet proven, and if the SaaS/API/usage-based transition stalls, the valuation logic would be fundamentally broken. Third, financial and working-capital risk: rising receivables and impairment are already a reality, not a distant worry. Fourth, dilution risk: low-strike options and RSUs with no performance conditions are not friendly to per-share value. Fifth, valuation risk: buying at the current price means bearing not so much "whether operations will collapse" as "whether the market is willing to keep telling its story at an extremely high multiple."

The strongest counter-view is, in fact, also the main view I currently hold: the market is not buying today's 51WORLD, but a 51WORLD that may one day become "China's physical-AI training infrastructure platform." But the statements disclosed today tell us that 51Aes still dominates group revenue, and gross margin collapsed significantly in 2025; while 51Sim has enormous room for imagination, it makes up only about 16% of group revenue; and the metrics that would truly prove the moat — high-margin recurring revenue, positive Owner Earnings, and a stable, replicable high ROIC — have so far not appeared. If over the next two or three years the company merely keeps "growing revenue, consuming cash, and diluting shares," then this investment could well degrade from a "high-expectation growth stock" into a "high-volatility value trap."

To make "being wrong" more concrete: if the company is re-rated by the market at 20–30x P/S rather than today's sales multiple of more than a hundred times, then based on 2025 revenue and net cash, the share price would roughly fall into the HK$20–29 region, still about 70%–80% of downside from the current level; and if, on top of that, gross margin fails to return to 40% for a long time, the recurring-revenue transition fails, and receivables keep deteriorating, long-term value could even move closer to the asset-approach range. Conversely, only when 51Sim truly expands from "narrow-niche leader" into "the group's main profit engine," and lifts group gross margin, cash flow, and per-share intrinsic value all together, would the cautious conclusion I hold today be overturned by facts.

Below is the checklist I provide under a long-term value-investing framework:

My final conclusion is as follows.

【Final Rating】 Avoid

【One-Sentence Investment Thesis】 51WORLD sits in a very good space, and 51Sim is full of imagination, but what you buy today is still a company with negative free cash flow, an unsettled moat, and a price set by the market on extremely high expectations.

【Core Bull Case】

• The digital-twin design-and-operations, AI training-and-validation, and physical-AI infrastructure spaces the company occupies still have relatively high industry growth in the coming years.

• 51Sim shows a relatively strong competitive position in the more narrowly defined end-to-end advanced autonomous-driving simulation and data platform market, and has built partnerships with leading automakers and testing institutions.

• The founder's shareholding and control are relatively strong, and management keeps directing resources into R&D rather than aggressive acquisitions.

• After listing the company has ample cash on hand, with no obvious problems in short-term solvency and survivability.

【Core Bear Case】

• In 2025 the group still lost RMB 186 million, both operating cash flow and free cash flow were negative, and Owner Earnings have not turned positive.

• Current revenue is still clearly project-leaning, with 98.6% of revenue recognized at a point in time, and repeatable, predictable subscription-type revenue has not yet been proven by the statements.

• In 2025 gross margin fell sharply from 51.1% to 30.0%, showing clear volatility in business quality.

• Receivables and impairment provisions rose significantly, and cash-profit quality is weak.

• The option and RSU arrangements are unfriendly to minority shareholders, with ongoing dilution risk.

【Key Assumptions】

• 51Sim must become a larger source of group revenue and profit over the next three to five years, not merely a valuation narrative.

• Group gross margin needs to recover back above 40%, otherwise the high-quality software-platform logic is hard to sustain.

• Operating cash flow must improve markedly and ultimately turn positive within the next two to three years.

• Standardized/SaaS/API usage-based revenue needs to genuinely raise its share.

• Equity dilution must be lower than the growth rate of per-share intrinsic value.

【Fair Buy Price】 HK$5–10/share. This is the price band after building in a sufficient margin of safety on top of the fair value range I provide.

【Target Holding Period】 If the business model is validated in the future, the holding period should be measured in 5–10 years or more; but at the current price and current financial quality, I do not recommend establishing a long-term position.

【Expected Annualized Return】 Buying at the current price and holding for 10 years, the rough annualized returns my model gives are:

• Conservative scenario: about -21%

• Base scenario: about -8%

• Optimistic scenario: about +3% These are not precise forecasts, but results derived from the current price, share count, 2025 revenue, and net cash as a starting point, under different growth and margin assumptions.

【Maximum Loss Risk】 If the valuation falls back to the 20–30x P/S range of comparable growth software/data companies, the share price could roughly return to HK$20–29; if the business-model transition fails and the market prices it by the asset approach rather than the growth approach, the long-term decline could be even larger. For a buyer at the current price, a 70%–90% permanent loss of capital is not unimaginable.

【Tracking Metrics】

• 51Sim revenue as a share of group revenue

• Standardized/SaaS/API subscription or usage-based revenue as a share of the total

• Whether group gross margin can return above 40%

• Whether operating cash flow and free cash flow turn positive

• Changes in receivables, aging structure, and impairment provisions

• Total share count, unexercised options, and RSU progress

• Whether partnerships with leading automakers turn into real recognizable revenue

• Whether the share of overseas revenue genuinely rises

• The R&D expense ratio and the efficiency of converting R&D into results

• Changes in period-end net cash and bank-loan structure

【Signals to Trigger Re-evaluation】

• By around 2027, the share of recurring revenue still shows no clear improvement

• 51Sim continues to be "strong story, weak statements"

• Gross margin stays below 35% for a long time

• Operating cash flow stays negative with no improving trend

• Receivables and impairment keep growing faster than revenue

• Management further expands low-cost equity incentives, clearly eroding per-share value

• Competition under the broad market definition intensifies, and the narrow-niche share advantage cannot be stably maintained

【Final Recommendation】 For an investor who is balanced-to-conservative with an investment horizon of more than 10 years, the most reasonable place for 51WORLD now is not "buy" but the watch list, waiting for two things to appear: first, a genuine transition of the business model from project-based to high-quality recurring revenue; second, a price that returns to a sufficiently conservative range. Until then, it looks more like a name worth studying but not worth betting long-term core capital on.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.