Quick ReadPlain-language overview · read this first
Tongfu Microelectronics is a major chip packaging and testing company in mainland China, with a research rating of "Watch." Through its two joint-venture plants in Suzhou and Penang, it is deeply tied to AMD and holds more than 80% share of AMD's related products. AMD's cycle directly determines its order elasticity, while the profit driver is shifting from traditional packaging and testing toward advanced packaging for AI and high-performance computing.
The key fundamental issue is the quality of earnings: revenue reached a record high in 2025, and net profit attributable to the parent was 1.219 billion yuan, but recurring net profit was only 841 million yuan. The gap mainly came from investment income and fair-value changes. In the first quarter of 2026, the gap widened further, while operating cash flow declined year on year. The core business improvement is real, but reported profit overstates its purity.
The moat lies in customer switching costs: changing packaging and testing suppliers for high-performance chips affects yield and delivery cadence, making replacement difficult once the relationship is embedded. Ultra-large multi-chip co-packaging has already entered mass production, giving the technology upgrade a concrete anchor. The flip side of this binding is dependence. TSMC and ASE are increasing investment in advanced packaging, and front-end giants' in-house operations remain the ceiling for the entire industry.
Valuation is the research report's biggest reservation: the current price is 60.10 yuan, with a P/E ratio of about 76 times based on reported profit and more than 110 times based on recurring profit, clearly above global peers. The ideal buying range is 28 to 36 yuan. The current price is far above that range and above the upper end of the holdable zone, leaving zero margin of safety. The report believes it is worth waiting, on the condition that the price falls back below 36 yuan and the recurring profit margin continues to rise.
There are three main risks: in 2024, revenue from a single customer accounted for 50.35%, so AMD order diversion or a slowdown would pressure profit first; U.S. restrictions on AMD's AI chip exports to China would transmit along the supply chain; if capacity expansion materializes slowly, depreciation and dilution would arrive before profit. The report's conclusion is Watch: the AMD chain is delivering quickly, but recurring profitability and margin of safety have not yet caught up with the share price. The above is a summary of the research report's views and does not constitute investment advice. The stock market involves risks; invest with caution.
LeadA major mainland China OSAT provider, Tongfu is deeply tied to AMD through its Suzhou and Penang joint-venture plants, accounting for more than 80% of AMD's related products. Revenue reached RMB 27.921 billion in 2025 and net profit attributable to shareholders reached a record RMB 1.219 billion, but recurring net profit was only RMB 841 million and recurring PE is above 110x. Research rating Watch: AMD-linked demand is materializing quickly, but recurring earnings and the margin of safety have not caught up with the share price, leaving the ideal buy range at RMB 28 to 36.
Prices in the article are as of publication; see the valuation band above for the live price.
Metadata
Ticker: 002156.SHE
Company name: Tongfu Microelectronics Co., Ltd.
Current price and market cap: RMB 60.10 / about RMB 93.241 billion, as of the close on 2026-06-11
Currency: CNY
Report date: 2026-06-11
Industry classification: Semiconductors
One-line positioning: A major mainland China OSAT provider that earns money through deep AMD ties and higher-end packaging upgrades.
This report was directly assigned by the operator from the zh.app "AI Supply Chain" topic. The research base date is 2026-06-11. The investment lens is comprehensive research, covering both a 12-month window and a 3 to 5 year observation window. Risk preference is treated as balanced. The analysis below is based mainly on the company's latest public disclosures, with AMD binding depth, the pace of advanced-packaging realization, earnings quality, and valuation margin of safety as the main axes.
Research Summary
On the surface, Tongfu Microelectronics is an OSAT company. What has truly led the market to reprice it in this cycle is a more specific set of words: AMD, advanced packaging, and AI servers. Its earnings machine today is built on two TF-AMD joint-venture plants in Suzhou and Penang, multi-site capacity coordination across Nantong, Hefei, and Xiamen, and a heavy-asset platform with capabilities in high-end flip-chip, wafer-level, and multi-chip packaging. It is no longer merely a volume manufacturer dependent on traditional leadframe and standard packages. In 2025, revenue reached RMB 27.921 billion and net profit attributable to shareholders reached RMB 1.219 billion, both record highs. In Q1 2026, revenue rose again to RMB 7.482 billion and attributable net profit reached RMB 329 million, with an even sharper year-on-year increase. The issue is that reported profit is not the same as operating earnings. In 2025, investment income plus fair-value gains and losses alone accounted for close to 30% of total profit among non-recurring items. In Q1 2026, only RMB 172 million of the RMB 329 million in attributable net profit was recurring, while operating cash flow fell 35.4% year on year. The key analysis for this company is now which profits are repeatable and which are simply accounting amplifiers in a favorable tape, rather than whether the headline results look good.
The market's core trading narrative is also highly concentrated. The first layer is continued volume growth in AMD AI accelerators and server CPUs. The annual report states clearly that the "joint venture plus cooperation" model formed through acquisitions made Tongfu AMD's most important OSAT supplier, accounting for more than 80% of its related products. AMD itself generated record revenue of USD 34.6 billion in 2025, and Q1 2026 data-center revenue reached USD 5.8 billion, up 57% year on year, driven by continued ramp-up in EPYC and Instinct. The second layer is the advanced-packaging upgrade. Tongfu's 2025 annual report has moved beyond concepts, directly disclosing that ultra-large FCBGA multi-chip co-packaging entered volume production, Penang's 3nm multi-chip packaging passed validation, and bumping plus wafer testing were put into production in October 2025. The third layer is the expectation that domestic advanced-packaging demand can be absorbed locally. The company's 2026 refinancing projects expand capacity around wafer-level, automotive and other emerging applications, high-performance computing, and communications. Management also described the projects in investor communications as serving domestic substitution and key chip products. The problem is that these three narrative layers are not realizing at the same pace: AMD demand is the most tangible today, advanced packaging is being implemented, while domestic computing-power absorption still lacks sufficiently strong public order evidence.
The reasons the stock rose and fell in the past are also clear. In the 2023 industry trough, the company still generated RMB 22.269 billion of revenue, but attributable net profit fell to only RMB 169 million, so the market treated it as a standard heavy-asset cyclical stock. From 2024 onward, as the industry recovered, AMD momentum improved, and mid-to-high-end products took a larger share, attributable net profit rebounded to RMB 678 million. In 2025 it jumped further to RMB 1.219 billion. The share price rose within 52 weeks from RMB 22.90 to RMB 78.56, then pulled back to RMB 60.10 in June 2026. The essence of the rally was that the market rewrote it from a traditional OSAT manufacturer into a high-beta AMD plus AI advanced-packaging proxy. It did not suddenly discover that Tongfu had always been cheap. Once investors price it under the latter framework, valuation moves much faster than profit realization.
The most important current bull-bear debate centers on one paradox: AMD binding is both the engine and the risk. Bulls argue that customer concentration is not bad as long as the customer is growing and products are migrating toward higher added value; concentration can even improve resource allocation efficiency. Bears argue that the higher the single-customer share, the weaker the pricing power, and the more the company becomes an amplifier of the customer's cycle. The 2024 annual report clearly disclosed that revenue from one single customer accounted for 50.35% of total revenue. In 2025, the company no longer publicly disclosed a named revenue share, but still wrote in the annual report that it is AMD's most important OSAT supplier and accounts for more than 80% of AMD's related products. At the same time, TSMC sees CoWoS as the capacity bottleneck of the AI era and expects CoWoS capacity to grow at a CAGR above 80% from 2022 to 2027. ASE also expects its advanced-packaging business to double again in 2026. For Tongfu, this means orders may grow, but it also means stronger front-end foundries and larger global OSATs are both adding capacity in advanced packaging.
Combining fundamentals, competition, and valuation, Tongfu Microelectronics today looks more like a company being put by the market into a script-rewriting phase. It is neither a high-quality compounder nor merely a cyclical turnaround. The old base remains heavy-asset OSAT, but the earnings driver is shifting from traditional packaging-and-testing turnover to AI/HPC-related high-end products and advanced packaging. The stock price has partly believed the new story, while the financials have not yet fully proved its stability. My qualitative label is: in valuation reset. The reason is simple: performance improvement is real and technology upgrades are progressing, but sustainable earnings power has not yet caught up with the current valuation premium. The market has already priced in what the company may become.
Company Development History
From a Local OSAT Plant to a Trajectory Rewritten by AMD
Tongfu Microelectronics' early identity was that of a typical back-end factory in Chinese manufacturing, far from the AI-packaging concept familiar to today's capital markets. Its former name was Nantong Fujitsu Microelectronics Co., Ltd., later changed to Tongfu Microelectronics. The name change itself explains the company's origin: it initially expanded by relying on foreign technology and local manufacturing capabilities, first solving the scale-manufacturing problem of packaging and testing, rather than earning its living from a single IP or architecture like a chip-design company. Public primary materials disclose little founder mythology, but what matters more for today's research is that from the beginning it survived on factories, equipment, yield, delivery, and customer coordination. For that reason, every later upgrade climbed step by step along the manufacturing platform toward higher-value back-end links, rather than through a change of track.
What truly changed the company's fate was the acquisition, announced in 2015 and completed in April 2016, of 85% equity interests in each of AMD's Suzhou and Penang packaging-and-testing plants, not an ordinary capacity expansion. Tongfu's official disclosure put the total investment at USD 371 million. AMD retained 15%, and the two sides formed the TF-AMD joint-venture structure. AMD later confirmed in SEC filings that the transaction turned Suzhou TF-AMD and TF AMD MICROELECTRONICS (PENANG) into ATMP joint ventures of the two parties. For Tongfu, this was a step change in capability, far more than buying two plants. A company that had leaned toward standardized OSAT was forcibly connected to the global high-end processor back-end supply chain for CPUs, GPUs, and APUs. After the acquisition, Tongfu became a platform deeply meshed with a leading customer in assets, production lines, processes, yield, and introduction cadence. It was no longer merely an order-taking OSAT.
The past decade can be divided into five stages. The first was the traditional OSAT platform-building period, focused on building factories and stabilizing yield and delivery to major customers. The second was the A-share expansion period, when multi-location capacity broadened product lines and created a basic manufacturing network in Nantong, Hefei, Xiamen, and other places. The third was the post-2016 AMD acquisition integration period, when the company absorbed the Suzhou and Penang joint-venture plants while carrying goodwill, multinational operations, and customer-concentration risk. The fourth was the 2022 to 2023 industry downturn and capability accumulation period: 2022 revenue was RMB 21.429 billion and attributable net profit was RMB 502 million; 2023 revenue was RMB 22.269 billion, but attributable net profit fell to RMB 169 million. This shows that both traditional OSAT and high-end bound businesses are affected by the semiconductor cycle, and the high-depreciation, high-debt heavy-asset structure can compress profit to a thin level. The fifth stage is the AI/HPC reacceleration since 2024: 2024 revenue was RMB 23.882 billion and attributable net profit was RMB 678 million; 2025 revenue rose to RMB 27.921 billion and attributable net profit to RMB 1.219 billion. Profit elasticity began to truly reflect the overlap of customer momentum, high-end products, and operating leverage.
The transitions between stages are also representative. After the AMD plant acquisitions, the company has long carried acquisition goodwill. In the 2024 annual report, the auditor still listed the goodwill formed by this acquisition as a key audit matter, disclosing goodwill carrying value of RMB 1.151 billion at the end of 2024 with no impairment provision. This detail matters. On one hand, it shows the acquisition remains a core variable in the balance sheet and earnings assumptions. On the other hand, if AMD's order structure, gross margin, or capacity utilization changes persistently in the future, goodwill impairment will return to the center of risk. In other words, AMD binding has never brought revenue without cost.
The 2024 to 2025 reacceleration differs from the 2016 step change. In 2016, the company bought capability. In 2025, it used that capability. The 2025 annual report states plainly that the global semiconductor market was driven by structural demand from AI computing, automotive electronics, consumer electronics, and new energy; capacity utilization improved, and revenue from mid-to-high-end products increased significantly. At the same time, industry investments around the supply chain and upstream-downstream links brought solid investment income. More importantly, the annual report placed technology progress at implementation points: ultra-large FCBGA multi-chip co-packaging and extreme thermal-management solutions had completed development and entered volume production; the Penang plant's 3nm multi-chip packaging passed validation; and bumping plus wafer testing entered production in October 2025. The story at this stage has become whether what has been built can generate sustained profit, not whether it can be built.
Financial Longitudinal Review
| Period | Revenue | Attributable net profit | Recurring attributable net profit | Net operating cash flow | Main feature |
|---|---|---|---|---|---|
| 2022 | RMB 21.429 billion | RMB 502 million | — | — | Industry momentum began to weaken |
| 2023 | RMB 22.269 billion | RMB 169 million | — | — | Downcycle compressed profit |
| 2024 | RMB 23.882 billion | RMB 678 million | RMB 621 million | about RMB 3.877 billion† | Demand recovery and mid-to-high-end product repair |
| 2025 | RMB 27.921 billion | RMB 1.219 billion | RMB 841 million | RMB 6.966 billion | AI/HPC pull and investment income uplift |
| 2026Q1 | RMB 7.482 billion | RMB 329 million | RMB 172 million | RMB 942 million | Reported profit clearly stronger than recurring earnings and cash flow |
† 2024 net operating cash flow is inferred from the 2025 annual report disclosure that 2025 grew 79.66% year on year. Revenue and attributable net profit for 2022 to 2024 come from the company's January 2026 investor-relations activity record. 2024 recurring profit comes from broker data compiled from the annual report. 2025 and 2026Q1 data come from the annual report and first-quarter report.
The most important point in this sequence is the change in profit structure, not the revenue growth rate. Revenue of RMB 22.269 billion in 2023 left only RMB 169 million of attributable net profit, showing that in a semiconductor downturn, an OSAT business with high fixed costs, depreciation, and finance expenses is extremely sensitive to utilization. Attributable net profit returned to RMB 678 million in 2024 and jumped to RMB 1.219 billion in 2025. Three forces worked together behind this: industry recovery, AMD and high-end product mix expansion, and tighter internal cost control. The annual report also emphasized overseas revenue of RMB 18.594 billion and domestic revenue of RMB 9.327 billion, with overseas revenue still high as a share. The company is clearly pulled by global customer momentum, not sustained by a purely domestic loop.
But looking only at attributable net profit overstates the purity of operating improvement. The 2025 annual report separately analyzed non-core business items: investment income was RMB 279 million, accounting for 15.93% of total profit; fair-value gains and losses were RMB 242 million, accounting for 13.79% of total profit, and the company explicitly marked them as not sustainable. In other words, at least part of the 2025 reported profit high came from paper gains on industry investments rather than the core packaging-and-testing business. Recurring net profit of RMB 841 million is also solid, but the valuation under that measure is much more expensive than under the reported measure. By Q1 2026, this structural issue was more obvious: attributable net profit was RMB 329 million, recurring profit only RMB 172 million, with the gap mainly from fair-value gains of RMB 183 million. Net operating cash flow was RMB 942 million and still positive, but down about 35.4% from RMB 1.458 billion in the prior-year period. This means the company is a typical case where core-business improvement is real and investment-income amplification is also real, rather than a case of fake high profit. Research must look at both measures side by side.
The balance sheet also explains why it can never be valued like a light-asset design company. At the end of 2025, cash and cash equivalents were RMB 5.410 billion, short-term borrowings RMB 3.489 billion, and long-term borrowings RMB 9.726 billion. Interest-bearing borrowings alone exceeded RMB 13.2 billion. Fixed assets were RMB 23.357 billion, accounting for 49.42% of total assets. This is a typical equipment-intensive OSAT structure. The advantage is large operating leverage when demand recovers. The drawback is that if new-line ramp-up is slow or utilization falls, depreciation will pressure the income statement first. For Tongfu, capex is the ticket to defend customer position. It has never been optional.
Share Price and Valuation History
Tongfu Microelectronics' share price over the past three years has amplified almost in sync with financial performance. In 2023, profit was extremely weak, and the market was willing to treat it only as an OSAT still alive in the cycle. In 2024 profit recovered, and in 2025 it set another record. The market began treating it as a China packaging proxy for AMD. In Q1 2026, profit kept growing sharply, while advanced packaging, domestic substitution, and refinancing-driven capacity expansion expectations pushed the stock from RMB 22.90 to a 52-week high of RMB 78.56 before it retreated to RMB 60.10. The pullback came because investors at high levels began distinguishing between reported profit and sustainable profit, not because the market rejected the AI logic.
At the 2026-06-11 closing price of RMB 60.10, the company traded at about 76x 2025 reported PE. On 2025 recurring net profit, PE exceeded 110x. Based on 2025 net assets per share of RMB 10.21, PB was about 5.9x. This valuation has already priced in continued AMD orders in 2026 to 2027, successful advanced-packaging upgrades, and ongoing margin improvement. It has clearly moved beyond pure cyclical pricing for a traditional OSAT. That is why Tongfu today is more worth studying for whether its valuation center has moved up ahead of operating quality.
Business Model and Moat
How It Actually Makes Money
Tongfu Microelectronics' revenue essentially comes from one-stop packaging-and-testing services: design simulation, wafer probing, packaging, final testing, and system-level testing. The technology page on the company's official website shows a broad package portfolio, from traditional leadframe packages and substrate-based packages to wafer-level and fan-out packages, covering consumer, industrial, automotive, high-performance computing, big-data storage, networking and communications, artificial intelligence, and other fields. The 2025 annual report further shows that what is truly lifting margins is mid-to-high-end products, especially product lines related to HPC, AI, automotive electronics, power management, storage, and display drivers, rather than low-end broad-based capacity. The company repeatedly emphasized in the annual report that revenue from mid-to-high-end products increased significantly. That line matters more than the revenue record because it explains why profit growth was clearly faster than revenue growth.
The cost side is highly typical: equipment, plants, engineering teams, quality systems, and customer validation are sunk costs; raw materials and part of labor are variable costs. Tongfu therefore has high operating leverage. When revenue rises, depreciation and factories do not grow at the same rate, so profit elasticity is amplified. When revenue falls, those same costs are hardest to cut, and profit is quickly squeezed. That is why 2023 revenue of RMB 22.269 billion generated only RMB 169 million of profit, while 2025 revenue rose to RMB 27.921 billion and profit jumped to RMB 1.219 billion. The difference was not just RMB 5.652 billion more revenue. It was the combined improvement in high-end products and capacity utilization, which diluted fixed costs.
The Real Moat and the Moat in Market Promotion
Tongfu's strongest moat is customer switching cost, not brand. The annual report is clear: the company and AMD have a post-acquisition "joint venture plus cooperation" model, far beyond a normal order-taking relationship. By 2025, the company remained AMD's most important OSAT supplier, accounting for more than 80% of its related products. Packaging and testing may look like outsourced manufacturing, but for high-performance chips such as CPUs, GPUs, and APUs, switching a back-end supplier is not like changing a screw supplier. Process windows, yield ramp-up, thermal management, substrate coordination, test programs, and delivery cadence all lock in the relationship. In favorable conditions, this barrier acts like an order amplifier. In adverse conditions, it acts like a customer-dependence amplifier. It is a real barrier, but not a comfortable one.
The second moat is technology and engineering accumulation. In the 2025 annual report, the company disclosed that ultra-large FCBGA multi-chip co-packaging had achieved volume production, while the Penang plant's 3nm multi-chip packaging passed validation and bumping plus wafer testing entered production. Earlier investor communications and broker materials show that the company has long laid out 2.5D/3D, Chiplet, Fan-out, and other areas, and had disclosed self-built line connection, 5-layer RDL large-size packaging solutions, and other capabilities. Here, verified and realized at scale must be separated. What the 2025 annual report confirms is 3nm multi-chip validation, FCBGA multi-chip volume production, and wafer-level capability expansion. Statements such as "5nm Chiplet has entered volume production" or "5-layer RDL ultra-large-size packaging has been widely commercialized" appear more often in 2022 to 2023 communications and sell-side research, lacking item-by-item confirmation in the 2025 annual report. This part is therefore better treated as evidence of technology reserve for now, not evidence of realized earnings.
The third moat is a global manufacturing network. The January 2026 investor-relations record shows that the company has nine major production bases in Nantong, Hefei, Xiamen, Suzhou, and Penang, Malaysia, with a global manufacturing and service network. For ordinary OSAT, this network is a scale advantage. For a multinational high-performance customer such as AMD, it is a compliance and delivery advantage. The Penang plant means Tongfu is not a line entirely locked into mainland China by geopolitics. But because its customer and technology chain is highly international, the company is also directly exposed to the transmission of U.S. export controls through AMD's business. Overseas bases are therefore both buffers and exposure points.
Management, Capital Allocation, and Governance
The current control and management structure is relatively stable. In the January 2026 investor-relations record, the company clearly stated that the major shareholder is Nantong Huada Microelectronics Group Co., Ltd., the actual controller is Shi Mingda, and the receiving executive was board secretary Jiang Shu. Audit opinions for the 2025 and 2024 annual reports were both standard unqualified opinions, with no public signal of modified audit opinions or major accounting disputes. The more relevant governance question is whether capital allocation is too aggressive, not whether the company will suffer an abrupt accounting event. In 2024, the company signed an agreement to acquire 26% of Greatek Electronics and completed delivery in February 2025. Management also said in investor communications that the transaction could improve investment income and bring stable financial returns. The investment did indeed lift profit, but it also deepened investor questions about the quality of core-business earnings.
On dividends, the company proposed a 2025 cash dividend of RMB 0.81 per 10 shares, materially higher than the prior year. But at a share price around RMB 60, the dividend yield is almost negligible. For a company like Tongfu, shareholder returns can only come from profit re-rating after successful capacity expansion, not stable dividends. That makes the standard for judging capital allocation simple: if new lines and refinancing projects truly produce high-end packaging revenue and recurring profit, today's heavy investment will be seen as necessary; if they only increase depreciation and dilution, the market will quickly rewrite growth investment as capex losing control.
Industry and Peer Analysis
Industry and Cycle Analysis
The OSAT industry is one of the easiest positions to misread in the semiconductor chain. It earns money from manufacturing execution, process coordination, validation capability, and customer stickiness. It lacks the extreme technology choke points of front-end manufacturing and the light-asset high margins of design companies. It naturally carries both inventory-cycle and capex-cycle attributes. According to Gartner data, the global integrated-circuit packaging-and-testing market was expected to reach USD 82.0 billion in 2024, up 7.8% year on year. At the same time, SIA/WSTS data show global semiconductor sales reached USD 791.7 billion in 2025 and may approach USD 975.0 billion in 2026. The industry base is still growing, but the fattest profit pools are moving toward advanced packaging, especially 2.5D/3D, CoWoS, FCBGA, and multi-chip heterogeneous integration tightly tied to AI servers and high-performance computing. Standardized packaging has not disappeared. It has become more of a scale competition.
For Tongfu, this industry stacks at least four cycles: the semiconductor cycle, the customer cycle, the capex cycle, and the technology-iteration cycle. The semiconductor cycle determines overall utilization. AMD's product cycle determines high-end order elasticity. Refinancing and equipment investment determine depreciation pressure. The pace from validation to volume production in advanced packaging determines whether margins can move up a level. In an upswing, the most beneficial variables are high-end order mix and capacity utilization. In a downturn, the most vulnerable variables are customer concentration and fixed costs. Whether Tongfu can pass through cycles depends on whether advanced packaging and non-AMD customers can absorb the lines in a downturn, not on whether a downturn occurs.
Policy and geopolitics must be analyzed separately in this cycle. The company wrote positively about the policy environment in the 2025 annual report, emphasizing support for industry-chain upgrades from 2025 tax-preference-list filings and domestic procurement standards. Management also described the 2026 refinancing projects as coordinating with downstream leading chip vendors, accelerating local capacity construction, and reshaping the supply system. The other side is that the United States imposed export-license requirements on AMD's MI308 and other AI chips for the China market. AMD itself listed restricted China sales as a risk factor in its 2025 10-K and estimated in April 2025 that related restrictions could bring up to USD 800 million of charges. For Tongfu, policy and geopolitics are not one-way positives or one-way negatives. Domestic substitution provides a story and opportunity, while external restrictions can cut part of the demand related to AMD's China market.
Horizontal Peer Analysis
Start with the most direct A-share peers. JCET is larger, with 2025 revenue of RMB 38.87 billion and attributable net profit of RMB 1.57 billion. Huatian Technology is smaller, with 2025 revenue of RMB 17.214 billion and attributable net profit of RMB 711 million. Around 2026-06-11, JCET's market cap was about RMB 127.7 billion, Huatian's about RMB 58.6 billion, and Tongfu's about RMB 93.2 billion. On the surface, all three had high TTM PE: JCET about 78x, Huatian about 68x, and Tongfu about 76x on reported net profit. But if Tongfu's recurring profit is used, valuation immediately jumps above 110x. In other words, among the three major A-share OSAT companies, the market has not given Tongfu a discount for its smaller scale. It has instead given it a clear elasticity premium.
| Company | 2025 revenue | 2025 attributable net profit | Market cap near research base date | Reference TTM PE |
|---|---|---|---|---|
| Tongfu Microelectronics | RMB 27.921 billion | RMB 1.219 billion | about RMB 93.2 billion | about 76x |
| JCET | RMB 38.87 billion | RMB 1.57 billion | about RMB 127.7 billion | about 78x |
| Huatian Technology | RMB 17.214 billion | RMB 711 million | about RMB 58.6 billion | about 68x |
Data sources: 2025 annual reports or annual-report disclosures of Tongfu Microelectronics, JCET, and Huatian Technology. Market cap and PE use public quotes around 2026-06-11.
These three companies have become three different models. JCET looks more like a platform-scale leader, with the market focusing mainly on scale, platform, and industry beta. Huatian looks more like a broad-based OSAT player with multiple layouts in automotive, RF, and storage. According to its investor records, it has bases in Xi'an, Jiangsu, and Malaysia's Unisem, laying out TSV, Bumping, WLCSP, Fan-Out, and other products, with a more diversified business mix. Tongfu looks most like a high-beta single-customer and high-end-product linkage model. Why do customers choose Tongfu? Because in certain high-performance product lines, it is deeply embedded in AMD's supply chain, and the Suzhou and Penang joint-venture plants turn that binding from a commercial relationship into a production-line relationship. It has never relied on being the most comprehensive. Why might customers also leave it? Because truly high-end advanced packaging never depends only on one back-end supplier. Front-end capacity, substrates, testing, and system coordination are all being consolidated again. As customers move toward the leading edge, they may hand more packaging to front-end giants such as TSMC or diversify among larger OSATs such as ASE and Amkor.
Globally, Tongfu's two real threats are asymmetric. ASE is the scale threat. Reuters cited its earnings call as saying ASE expects advanced-packaging revenue to double again in 2026 to USD 3.2 billion and will continue aggressive capex. This means the world's largest OSAT is treating AI packaging as the next growth axis. Amkor is the valuation-reference threat. Its current share price was USD 69.61, market cap about USD 17.37 billion, and PE about 40x, clearly below Tongfu's 76x on reported earnings and far below more than 110x on recurring earnings. In other words, global markets are willing to pay for advanced-packaging elasticity, but usually not a sustained premium like Tongfu's today under such high customer concentration. The third threat is more structural: TSMC. In its 2025 annual report and public comments in May 2026, TSMC emphasized strong CoWoS demand and a CoWoS capacity CAGR above 80% from 2022 to 2027. Front-end foundries doing more advanced packaging themselves is the ceiling risk for all OSATs.
Therefore, Tongfu's position in the industry today is closer to a challenger pulled into the high-end track by one super customer. It is neither a leader nor a pure follower. The gap it fills is that among mainland China OSATs, it is one of the few players able to combine high-performance CPU/GPU/APU back-end work, large-size FC, wafer-level capabilities, and overseas-base delivery. The orders it most directly competes for are mainland high-end packaging-and-testing orders and part of global OSAT share. The profit pool most likely to be taken from it is threatened by TSMC's in-house advanced packaging, ASE's scale advantage, and AMD's own supply-chain diversification.
Current Fundamentals, Valuation, and Risks
Current Fundamentals and Bull-Bear Debate
Over the latest four quarters, Tongfu's operating state has broadly been rising revenue, larger profit elasticity, but earnings quality that has not fully moved in sync. Full-year 2025 revenue was RMB 27.921 billion and attributable net profit RMB 1.219 billion. Q4 revenue alone was RMB 7.81 billion and attributable net profit RMB 358 million. Q1 2026 revenue rose again to RMB 7.482 billion and attributable net profit to RMB 329 million. The company's explanation for the Q1 change was direct: operating revenue increased year on year, especially because revenue from mid-to-high-end products increased significantly, while industry investments around the supply chain and upstream-downstream layouts produced solid returns. This explanation does not avoid the issue. It shows the company also knows the market is watching not just how much growth there is, but where it comes from.
The market is mainly trading three clues. The first is AMD AI/HPC realization. AMD's record 2025 revenue and high Q1 2026 data-center growth give Tongfu the prospect of simultaneous uplift in order volume, ASP, and advanced-packaging content. The second is advanced-packaging refinancing and capacity expansion. In early 2026, the company launched a private A-share issuance plan to expand around wafer-level packaging and testing, automotive and other emerging applications, high-performance computing, and communications. In essence, it is telling the market that existing capacity is insufficient and that the company is willing to keep committing heavy capital to higher-end packaging. The third is the idea of domestic absorption. But the public evidence for this clue remains weak so far. The company has disclosed a domestic-substitution positioning and wrote in the Suzhou TF-AMD quality report that major customers include AMD and domestic "China chip" customers. Yet in public annual reports and quarterly communications, there is still not enough strong named-order and profit-contribution evidence. I therefore prefer to treat domestic AI-packaging absorption as an option rather than a realized fact.
What bulls truly see is a path that is already halfway through: AMD itself is performing strongly, the company's binding with AMD is not an ordinary outsourcing relationship, advanced packaging is moving from R&D and validation into volume production and production start-up, and the global OSAT industry is indeed enjoying structural incremental demand from AI. If these factors combine, the company's revenue and recurring-profit center may continue to move upward over the next 2 to 3 years. What bears focus on is the other side: the single-customer share was already as high as 50.35% in 2024; although 2025 no longer gave a named share, it still emphasized an 80%+ related-product share; 2025 and 2026Q1 profits were both clearly amplified by investment income and fair-value changes; TSMC and ASE are both expanding advanced packaging; and refinancing means depreciation, capex, and dilution pressure continue to rise. In other words, Tongfu's current question is how many fragile links are embedded in a good story, not whether there is a story.
Valuation Analysis
Tongfu's current valuation cannot be viewed through only one measure. Based on 2025 reported attributable net profit of RMB 1.219 billion and a market cap of about RMB 93.2 billion, PE is about 76x. Based on 2025 recurring attributable net profit of RMB 841 million, PE is about 111x. On PB, sell-side commentary on the 2025 annual report gave net assets per share of about RMB 10.21, corresponding to current PB of about 5.9x. For a heavy-asset OSAT, this clearly already prices in advanced-packaging realization over the next few years and is far beyond a valuation for recovery from a cyclical bottom.
After looking through cash flow, the picture becomes a little more complex. In 2025, net operating cash flow was RMB 6.966 billion, far above net profit, mainly because of depreciation, amortization, and the fixed-asset cycle. The 2025 annual report also clearly stated that the difference between operating cash flow and net profit was mainly due to factors such as fixed-asset depreciation. If using broker estimates based on the company's financial statements, 2025 depreciation and amortization were about RMB 3.473 billion and capex about RMB 6.211 billion. Maintenance capex is unlikely to be lower than depreciation and amortization, implying owner earnings roughly in the RMB 2.6 billion to RMB 3.5 billion range and an implied yield of about 2.8% to 3.8%. This looks much better than headline PE, but it still does not mean the stock is cheap. Advanced packaging is a heavy-asset industry with rapid technology iteration, so maintenance capex cannot be as low as in utilities. My conclusion is that reported PE overstates part of the expensiveness, but the cash-flow measure is still far from cheap.
Peer comparison makes the point clearer. Among A-share comparables, JCET and Huatian have TTM PE of about 78x and 68x respectively, showing that the whole mainland OSAT sector has had its valuation center lifted by advanced packaging plus AI. But globally, Amkor's current PE is about 40x and ASE's about 51x, making Tongfu look expensive. The market is willing to give Tongfu a higher premium mainly because it looks more than most peers like a high-beta mapping of AMD orders, not because it is safer than global leaders. The issue is that once such a premium is established, it becomes very dependent on new quarters continuing to deliver the combination of high growth and high-end packaging progress.
| Dimension | Bearish | Base | Bullish |
|---|---|---|---|
| Revenue/margin assumption | 2026-2027 revenue grows about 8%–10% per year, recurring net margin stays at 3%–3.3% | Revenue grows about 12%–15% per year, recurring net margin rises to 3.8%–4.2% | Revenue grows about 18%–22% per year, recurring net margin approaches 4.8%–5.2% |
| Cash-flow assumption | OCF remains positive, but new lines lift depreciation and capex, keeping cash returns weak | OCF rises with profit and new lines ramp smoothly | High-end orders and utilization rise together, with cash flow improving faster than profit |
| Valuation multiple assumption | Recurring PE 40–50x | Recurring PE 50–60x | Recurring PE 60–75x |
| Key catalyst | AMD orders do not stall; Q1-style investment-income noise declines | Penang/Suzhou high-end lines ramp smoothly; non-AMD customers increase | AMD AI and EPYC continue to beat expectations; advanced packaging scales in area |
| Key risk | Customer concentration keeps rising; new depreciation pressures profit | Core-business profit improvement misses expectations after investment income fades | TSMC/ASE/other OSATs divert advanced-packaging orders |
| Implied share-price range | RMB 28–36 | RMB 42–55 | RMB 70–90 |
| Implied return from current price | -40% to -53% | -8% to -30% | +16% to +50% |
| Permanent-loss risk | Trigger: AMD share declines and expansion lands in a downcycle | Trigger: recurring margin fails to rise and valuation reverts first | Trigger: high valuation is overdrawn and any one link stalls |
The share-price ranges in the table use recurring earnings as the anchor, not reported net profit, because the company has already shown in its 2025 annual report and 2026Q1 that investment income and fair-value changes can materially disturb accounting profit. Foreign-exchange translation uses 2026-06-11 rates of 1 USD=6.7774 CNY and 1 TWD=0.2137 CNY.
The margin-of-safety review is direct. First, the current price is at a significant premium to the conservative range, so the margin of safety is zero. Second, the most fragile assumption across the three scenarios is whether the share of AMD-related high-end orders can keep rising and whether the resulting margin can stay with Tongfu rather than be captured by stronger supply-chain links. Overall industry demand is not the most fragile link. If that assumption is cut by 30%, the base-case share-price center would move from RMB 42–55 to about RMB 35–45. Third, if recurring profit shows zero growth over the next three years, the current valuation can almost only be maintained by sentiment, and high returns would not come from earnings themselves. As soon as the market is willing to view it again as a heavy-asset cyclical OSAT rather than an AI-beta proxy, today's entry price has no moat-like margin of safety. My independent conclusion is that margin-of-safety sufficiency is absent.
Risk Analysis
The first risk is that customer concentration keeps rising rather than falling. Probability is medium-high, and impact is high. In 2024, revenue from one single customer already accounted for 50.35%; the 2025 annual report still said AMD-related products accounted for more than 80% of AMD's related-product share. If AMD's own growth slows, or if it shifts higher-end packaging further toward TSMC, ASE, or other supply-chain partners, the income statement will feel pressure before the revenue statement. The most important metric to watch is whether single-customer share continues to rise in the next annual report and whether management still repeatedly emphasizes the "more than 80%" share, rather than quarterly revenue alone. The transmission path would be direct: order structure changes, advanced-product ASP and utilization decline, gross margin falls first, and high depreciation then amplifies the profit decline.
The second risk is the transmission of export controls to AMD's China business and related supply chain. Probability is medium, and impact is high. In its 2025 10-K, AMD explicitly listed restricted China sales as a risk factor and named the MI308 data-center GPU as affected by export controls. In April 2025, AMD also estimated that new restrictions could bring up to USD 800 million of charges. Tongfu does not directly sell AI chips, but the customer it is deeply bound to is being cut by regulation across markets. If later restrictions expand to China-linked entities outside China or tighten further, Tongfu's order cadence, product mix, and even valuation narrative would be affected. Observable signals for investors include AMD licensing progress for China-related products, inventory and charge recognition, and whether Tongfu further emphasizes domestic substitution and new-customer introduction in public communications as a hedge.
The third risk is that the success rate of advanced-packaging expansion is lower than the share price assumes. Probability is medium, and impact is high. The company's 2026 refinancing projects expand around wafer-level, high-performance computing and communications, automotive, and other directions, which itself shows management believes existing capacity needs to expand. But advanced packaging and automotive/high-performance products are not businesses where machines arrive and profit follows. They require customer introduction, validation, yield, substrates, and testing to work together. The annual report can confirm that 3nm multi-chip packaging passed validation and FCBGA multi-chip entered volume production, but significant profit contribution still requires the ramp-up step. If expansion meets demand slowdown or introduction misses expectations, depreciation and dilution arrive first and profit later. Variables to watch include the refinancing pace, fixed assets and construction-in-progress changes, and whether management's description of the mid-to-high-end product share keeps strengthening.
The fourth risk is that earnings quality continues to be masked by investment income. Probability is high, and impact is medium-high. In 2025, investment income plus fair-value gains and losses accounted for close to 30% of total profit. In 2026Q1, the gap between reported attributable net profit and recurring profit was about RMB 157 million, while fair-value gains were RMB 183 million. As long as this continues, the market will swing repeatedly between the core business getting stronger and profit not all coming from the core business. It will not immediately hurt revenue, but it will hurt valuation quality. A practical tracking rule is that if non-recurring gains and losses continue to account for more than 30% of attributable profit over the next two quarters and recurring margin does not rise materially, the market will probably compress valuation.
The fifth risk is that valuation itself has already pulled too much forward. Probability is high, and impact is high. At the current price, Tongfu trades at about 76x PE on reported net profit and more than 110x on recurring net profit. Amkor's current PE is about 40x and ASE's about 51x. Unless the company can keep delivering simultaneous improvement in revenue growth, recurring profit growth, and advanced-packaging progress, the market has no reason to maintain the current premium for long. Valuation damage in such stocks often comes faster than earnings downgrades. The watch variables are clear: recurring profit, operating cash flow, customer concentration, and advanced-packaging volume-production descriptions. If two of these four deteriorate at the same time, the valuation center may step down first.
Catalysts and Tracking Indicators
There are four main positive catalysts. First, AMD's data-center business continues to beat expectations, and high growth in EPYC and Instinct transmits to Tongfu's high-end orders in Suzhou and Penang. Second, in the next financial report, recurring profit and operating cash flow improve together, reducing market concerns about earnings quality. Third, Penang 3nm multi-chip packaging and ultra-large FCBGA multi-chip co-packaging move from validation and volume production toward clearer revenue contribution. Fourth, refinancing projects progress smoothly, and management can provide more specific clues on high-end packaging customers and capacity utilization.
Negative catalysts are equally clear. First, export restrictions on AMD products for China expand further, or China-related product shipments are blocked again. Second, quarterly reports again show the combination of strong reported profit, ordinary recurring profit, and weakening cash flow. Third, spillover from TSMC CoWoS or global OSAT leaders' advanced-packaging capacity appears faster, leading the market to question Tongfu's share ceiling. Fourth, refinancing-driven expansion progresses, but revenue growth and capacity utilization do not realize in sync.
| Indicator | Latest reference | Normal range | Warning threshold | Main tracking source |
|---|---|---|---|---|
| Single-customer revenue share | 50.35% in 2024 | Does not keep rising | Rises again in the next annual report to around 55% or higher | Annual-report notes |
| AMD-related product share | 2025 annual report says 80%+ | Maintained or lower with more diversified customers | Annual report stops emphasizing it and no new customers fill the gap | Annual report/IR |
| Recurring attributable net margin | about 3.0% in 2025, about 2.3% in 2026Q1 | Above 3%–4% | Below 2.5% for two consecutive quarters | Quarterly reports |
| Operating cash flow / attributable net profit | about 5.7x in 2025, about 2.9x in 2026Q1 | Above 1 | Below 1 for two consecutive reporting periods | Annual/quarterly reports |
| Non-recurring gain/loss disturbance | about 29.7% of total profit in 2025 | Below 20% | Above 30% and persistent | Annual/quarterly reports |
| Advanced-packaging progress | 3nm validation, FCBGA multi-chip volume production | Continued scale-up after validation | Stays at validation without translating into revenue | Annual report/IR |
| Refinancing and capex pace | 2026 private-placement projects in progress | Financing moves in sync with orders | Capex rises but orders do not follow | Announcements/IR |
The thresholds in the table are meant to turn story realization into trackable hard conditions. They are not trading signals. The most important combination is customer concentration, recurring margin, and advanced-packaging progress, not any single indicator. If all three improve in the same direction, Tongfu can continue to enjoy valuation reset. If two of the three weaken, the share price will often react first.
Combined Longitudinal and Horizontal Conclusion
Longitudinally, Tongfu Microelectronics has truly proved that it can keep lifting a manufacturing platform one level higher along customer demand, rather than that it can always grow at a high rate. From traditional OSAT origins, to cutting into the CPU/GPU/APU supply chain through the acquisition of two AMD plants, to putting FCBGA, wafer-level capabilities, and 3nm multi-chip validation into the annual report today, this company has used many years of heavy assets, slow ramp-up, and strong execution to push itself from an ordinary back-end plant into the high-performance computing chain. It did not rely on one revolutionary transformation. Its past success includes the wind of the industry cycle and management's willingness to bet at key moments, but the most important factor was still the 2016 AMD acquisition. Without that acquisition, today's market would not discuss it as an AI advanced-packaging elasticity stock.
Horizontally, its largest real advantage relative to peers is symbiotic binding with AMD and the production-line, validation, and delivery coordination formed behind it. Its largest real weakness is the same point. Compared with JCET, Tongfu is smaller and has a thinner cushion. Compared with Huatian Technology, it has stronger high-performance computing attributes but higher customer concentration. Compared with ASE and TSMC, it lacks the thickness of an industry leader that can still earn money from the platform even after losing one customer. Tongfu's growth is real, but the share price has priced in too much of that growth too early. This does not deny the company is becoming stronger. It is a reminder that when a company is re-rated at the same time by one super customer, one high-momentum theme, and one large expansion round, the market is most likely to misjudge its replicability.
There are two places where the market is now most likely to be wrong. The first mistake is linearly extrapolating AMD's high growth into Tongfu enjoying margin uplift for the next several years. Continued AMD growth is certainly positive for Tongfu, but the most valuable link in the supply chain may not always sit with Tongfu. The second mistake is treating the leap in reported profit as proof that core-business earnings power has already stabilized at a higher level. The 2025 and 2026Q1 data clearly show that investment income and fair-value changes materially amplify profit. What truly determines the valuation ceiling is whether advanced-packaging profit can stand on its own after investment income is removed, not how large the AMD story is.
The most important variable over the next year is whether recurring profit and operating cash flow can keep improving together. The key variables over the next three years are utilization after advanced-packaging expansion and whether non-AMD customers can fill in. The key variable over the next five years is whether the company can evolve from a high-beta AMD-bound proxy into a high-end packaging platform company. If it can only remain in the first position, Tongfu will be rewarded cyclically and punished cyclically. Only by reaching the second position does its valuation center earn the right to step up. Based on current evidence, my conclusion is more conservative: the company is worth continued tracking, but not worth rushing to own at today's price.
Bull and Bear Cases
The bull case can be summarized in five points. First, AMD achieved record revenue in 2025 and Q1 2026 data-center revenue kept growing strongly, while the company still accounts for more than 80% of AMD's related packaging-and-testing share, so order elasticity is real. Second, the 2025 annual report disclosed FCBGA multi-chip co-packaging volume production, Penang 3nm multi-chip validation, and bumping/wafer testing production start-up, so advanced packaging is not just a slide deck. Third, revenue and profit reached consecutive highs in 2024 to 2025, showing that mid-to-high-end product upgrades and operating leverage are being realized. Fourth, refinancing projects expand around wafer-level, high-performance computing and communications, automotive, and other directions, and management is still betting along the high-added-value path. Fifth, operating cash flow has long been materially above net profit, showing that a high-depreciation manufacturing platform can generate strong cash in an upcycle.
The bear case must be equally specific. First, 2024 single-customer revenue share was already as high as 50.35%, making customer-concentration risk a fact, not an imagination. Second, 2025 and 2026Q1 profit were both clearly thickened by investment income and fair-value changes, making reported profit less sustainable than the surface numbers suggest. Third, TSMC's in-house CoWoS and the expansion of ASE and other global OSATs are simultaneously squeezing the advanced-packaging profit pool. Fourth, Tongfu's current reported PE is about 76x and recurring PE exceeds 110x, clearly higher than global comparable Amkor, leaving insufficient margin of safety. Fifth, the company still needs to continue expansion and refinancing in 2026. If demand and yield realization lag even slightly, depreciation, dilution, and cash-flow pressure will all emerge together.
Pre-mortem
If this investment loses 50% three years from now, the most likely first scenario is as follows: in 2027, AMD accelerates diversification in advanced packaging, keeping more high-end product back-end work within the TSMC system or diverting it to ASE and other OSATs. Utilization of Tongfu's high-end lines in Suzhou and Penang falls from high levels to around 70%, gross margin declines by 2 to 3 percentage points, and recurring net profit returns to the RMB 800 million to RMB 900 million level. The market reprices it from a high-beta AI-chain proxy into a heavy-asset cyclical OSAT, compressing valuation from 50–60x forward PE back to 30–35x. That creates a path to a 50% share-price decline. This scenario does not require AMD's business to collapse. It only requires some supply-chain share movement and some valuation reversion. Its real-world basis is today's already high concentration and TSMC plus ASE's significant advanced-packaging capacity increases.
The second scenario is that in 2027 to 2028, Tongfu's expansion lands in the latter half of a high-cycle period. Private-placement projects keep advancing, fixed assets and depreciation rise further, but public disclosures still lack sufficiently strong new-customer and volume-production realization. After non-recurring gains and losses fade, real core-business profit does not rise as quickly as the market expected. Operating cash flow begins to lag new investment, and the market discovers that it remains a heavy-asset manufacturer that temporarily benefited from AI orders. Valuation switches from growth premium back to cyclical discount. The damage in this scenario comes from today's valuation leaving no room for such a stall, not from an instant deterioration in the income statement.
Final Research Conclusion
Tongfu Microelectronics is a company ramping toward high-end packaging, and its 2025 to 2026 results have proved that it has indeed benefited from the AMD and AI/HPC cycle. It is not a pure concept stock. Advanced-packaging progress and high-end orders can be found in annual reports, quarterly reports, and investor communications. But it is also far from a simple growth stock where reported profit alone is enough. What deserves respect is that it has gradually turned a traditional OSAT plant into a named and positioned link in the global high-performance computing supply chain. What deserves caution is that this success remains highly dependent on one super customer and a still-expanding high-capex logic.
At today's price, my biggest concern is that growth realization may not cover the valuation's pull-forward of the future, rather than that the company lacks growth. As long as the market is willing to treat Tongfu as a dual proxy for AMD and mainland high-end packaging, it will remain expensive. But once the market more seriously compares recurring profit, cash flow, and global comparable valuations, that premium becomes fragile. The conditions that would change my view are also specific: in the next stage, I need to see not only continued revenue growth, but also stable recurring-margin improvement, a clear decline in the share of non-recurring gains and losses, advanced packaging moving from validation/volume production toward visible profit contribution, and single-customer concentration no longer worsening.
【Company Profile Score】
Fundamental quality: Medium
Growth: Medium
Moat: Medium
Financial resilience: Medium
Management credibility: Medium
Valuation attractiveness: Low
Risk level: High
Suitable investor type: Cyclical
【Investment Rating】
Rating: Watch
One-line investment thesis: The AMD chain is realizing quickly, but recurring earnings and the margin of safety have not caught up with the share price.
Three-tier price signals: Ideal buy price: 28–36 CNY
Holdable price: 42–55 CNY
Clearly overvalued price: above 70 CNY
Current price classification: Outside the three tiers
Worth waiting for a better price: Yes. More ideal buy triggers would be the share price falling below RMB 36 and the next stage of financial reports proving at the same time that recurring margin continues to rise and customer concentration no longer worsens. The opportunity cost of waiting is the possibility of missing a phase of valuation uplift driven by the AI theme.
Target holding period: 1–3 years
Expected annualized return: conservative scenario -18% to -30%; base scenario -3% to +6%; optimistic scenario +12% to +25%
Maximum loss risk: about 50%; triggers would be AMD order diversion, advanced-packaging expansion missing expectations, and valuation switching from high-beta growth stock back to heavy-asset cyclical stock.
Signals that trigger reassessment: If single-customer revenue share continues to rise in the next annual report and approaches or exceeds 55%
If recurring net margin stays below 2.5% for two consecutive reporting periods
If investment income and fair-value changes continue to account for more than 30% of total profit
If advanced-packaging progress remains stuck for a long time at validation and volume production without translating into margin improvement
If AMD China-related export restrictions escalate materially again and affect product volume growth
【Ideal/Fair Buy Price】28–36 CNY Basis: This corresponds to the conservative scenario range of RMB 28–36, already incorporating customer-concentration risk, expansion uncertainty, and valuation-reversion pressure, while preserving a margin of safety of at least about 20% versus a relatively conservative value.
【Valuation Range】
current: 60.10, as of the close on 2026-06-11
bear, conservative ideal buy range: [28, 36]
base, reasonable acceptable holding range: [42, 55]
bull, optimistic and above the clear overvaluation line: [70, 90]
Key Data Table
| Key item | Latest value | Core meaning | Source |
|---|---|---|---|
| 2025 revenue | RMB 27.921 billion | Record high, with cycle and high-end products resonating | Annual report |
| 2025 attributable net profit | RMB 1.219 billion | Headline number is bright | Annual report |
| 2025 recurring attributable net profit | RMB 841 million | Closer to sustainable earnings | Annual-report summary |
| 2025 OCF | RMB 6.966 billion | Strong cash flow, but heavy-depreciation features are clear | Annual report |
| 2026Q1 attributable / recurring | RMB 329 million / RMB 172 million | Reported profit is materially stronger than operating earnings | First-quarter report |
| Single-customer share | 50.35% in 2024 | Hard anchor for customer-concentration risk | 2024 annual report |
| AMD-related product share | 2025 annual report says 80%+ | Deep binding, with growth and risk from the same source | 2025 annual report |
| 2026-06-11 closing price | RMB 60.10 | Already clearly prices in part of optimistic expectations | Market data |
| Reported PE / recurring PE | about 76x / 111x | Two profit measures lead to two conclusions | Calculated accordingly |
The "reported PE / recurring PE" in the table is calculated from the 2026-06-11 closing price and 2025 profit measures. It directly matches the core judgment of this report: Tongfu's debate has always been which profit measure should price growth, not whether there is growth.
Research Uncertainty
This report still has four important blind spots. First, the 2025 annual report did not continue to disclose the named single-customer revenue share as the 2024 annual report did. Therefore, regarding AMD's latest revenue share, I can only confirm the 80%+ related-product share, not give the audit-note-level exact number available for 2024. Second, the company's public disclosures on mass-production status for 2.5D/3D, Chiplet, Fan-out, and other advanced packaging use different expressions across years and occasions, including validation, line connection, volume production, and ramp-up. This report treats only the parts clearly confirmed by the 2025 annual report as realized facts. Third, the market narrative around domestic AI-chip packaging absorption is hot, but public order evidence was still insufficient as of the research base date, so this report applies a significant discount to it. Fourth, the final registration, fundraising cadence, and production schedule of refinancing projects will still change the medium-term earnings path.
Reference Sources
Full 2025 annual report, full 2024 annual report, January 2026 investor-relations activity record, company technology pages, and Suzhou TF-AMD quality credit report of Tongfu Microelectronics Co., Ltd.
Tongfu Microelectronics 2026 first-quarter report, 2025 annual-report summary, and related public interpretations.
AMD 2025 annual results, Q1 2026 results, 2025 10-K, and export-control-related disclosures.
Public data from WSTS, SIA, and Gartner on global semiconductor and OSAT market size.
TSMC 2025 annual report and Reuters coverage of CoWoS expansion.
Annual reports and performance materials of ASE, JCET, and Huatian Technology, plus current public quotes for Amkor.
Market prices and exchange-rate references as of the research base date: Reuters, Investing, Google Finance, Yahoo Finance/WISE/XE.
Other Securities Mentioned in the Report
AMD.US — Core customer, determining Tongfu's AI/HPC order elasticity and export-control transmission.
600584.SHG — JCET, the most important scale OSAT comparison in A-shares.
002185.SHE — Huatian Technology, a mainland third-tier leader with more diversified business and relatively lower customer concentration.
ASX.US — ASE Technology, the world's largest OSAT, whose advanced-packaging expansion is competitive pressure above Tongfu.
AMKR.US — Amkor, the valuation anchor for global comparable OSATs, with current PE clearly below Tongfu's.
TSM.US — TSMC ADR, whose in-house CoWoS/SoIC advanced packaging is Tongfu's strongest cross-boundary diversion threat.
This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.
Full report
Sign in to read the full report
Sign up free to unlock the full text, the Baillie growth scorecard, and full-text search.
Log in / Sign up free