Goldman Sachs says Kioxia's substantial capex does not preclude shareholder returns, while sustained buybacks could drive valuation expansion

Overview

The report analyzes Kioxia's newly announced Kitakami K3 fab building and medium-term capex plan and discusses whether these investments will constrain shareholder returns. Goldman Sachs concludes that the three-year average capex plan has not changed and that strong earnings and free cash flow could still support expanded buybacks beginning in FY3/28; combined with tight NAND supply-demand conditions, this could drive an increase in the valuation multiple currently depressed by cyclical concerns.

Core views

Goldman Sachs first notes that Kioxia's forecast P/E for FY3/28 is slightly above 3x, while global memory peers also trade at only low- to mid-single-digit P/E multiples. This is broadly consistent with valuations historically assigned at cyclical peaks and reflects widespread investor skepticism about whether current margins can be sustained. The report argues that the key to valuation upside may not merely be continued earnings growth, but whether the company can consistently return cash to shareholders. Kioxia recently conducted approximately ¥800bn in share buybacks; Goldman Sachs further assumes that, beginning in FY3/28, the company will define cash exceeding the equivalent of four months of sales as surplus cash and use most surplus free cash flow for shareholder returns, primarily through buybacks. Based on the current market capitalization, the cumulative shareholder return over FY3/28—FY3/29 is projected to approach 20%. The report uses Japanese shipping stocks as a historical analogy: after container freight rates surged during the pandemic, the market became concerned about the sustainability of earnings, and the P/E ratios of relevant shipping stocks once fell to 2—3x; however, even as the market subsequently weakened, proactive shareholder returns such as sustained buybacks continued to drive increases in share prices and valuation multiples. Goldman Sachs believes memory stocks currently face similar concerns about a cyclical peak, meaning stable and sustained capital returns could become a catalyst for the market to reprice Kioxia. Regarding capex, Kioxia announced plans for the new K3 fab building at its Kitakami plant after the market close on August 27, with operations expected to begin in FY3/30 and the specific production ramp-up schedule to be determined according to market trends. Kioxia and joint-venture partner SanDisk simultaneously announced plans to invest approximately ¥5tn in Japan through 2032, covering the six years from FY2026 (FY3/27) to FY2031 (FY3/32). The plan includes some R&D expenses, but most spending will go toward fab buildings, facilities, and equipment; the investment targets front-end production and excludes back-end processes. Based on historical investment proportions, Kioxia and SanDisk may bear 60% and 40%, respectively. Investment in the K3 fab building is approximately ¥1.8tn, while the overall plan also includes the Y7 fab building at the Yokkaichi plant. Despite the large scale of the long-term investment, Goldman Sachs believes near-term cash flow pressure has not clearly exceeded the company's existing framework. The average annual capex of approximately ¥470bn over FY3/27—FY3/29 presented by the company at its June IR Day remains unchanged, although the latest decision may bring some FY3/27 and FY3/28 investment forward slightly. Accordingly, the report believes the announced capex does not preclude expanded shareholder returns beginning in FY3/28, and capital investment and buybacks can proceed in parallel amid strong cash flow. Regarding earnings forecasts, Goldman Sachs modestly raises its operating profit forecasts for FY3/27—FY3/29 based on research conducted after first-quarter results and current operating conditions. The new operating profit forecasts are ¥8,426.6bn, ¥11,381.4bn, and ¥13,578.5bn, respectively, versus previous forecasts of ¥8,286.5bn, ¥11,371.8bn, and ¥13,463.9bn; the FY3/28 and FY3/29 forecasts remain approximately 10%—15% above Bloomberg consensus. The corresponding new EPS forecasts are ¥11,011.9, ¥16,476.9, and ¥24,167.5, respectively, versus previous forecasts of ¥10,827.8, ¥14,918.7, and ¥17,669.1. The corresponding forecast P/E falls from 4.8x in FY3/27 to 3.2x in FY3/28 and 2.2x in FY3/29. The forecast revisions do not stem from more aggressive shipment volume or pricing assumptions. As the sales mix shifts toward enterprise SSDs with longer delivery cycles, Goldman Sachs lowers its bit shipment forecasts; meanwhile, factors such as smooth progress in the transition to eighth-generation BiS make cost assumptions more favorable for earnings. Average selling price assumptions are not materially changed, and the report expects prices to continue rising moderately quarter over quarter until at least mid-CY27. The medium-term industry outlook remains positive. Kioxia is the world's third-largest NAND flash memory manufacturer, and Goldman Sachs is constructive on the higher margins enabled by its cost competitiveness and the company's gradual progress in the high-growth data center product segment. Rising demand for enterprise SSDs used in servers, tight supply of HDDs that are also used for storage, and market concerns that the United States may restrict exports of production equipment to Korean memory fabs in China are collectively tightening NAND supply-demand conditions. The report believes strong AI demand, limited near-term incremental supply, continued prioritization of DRAM investment by major memory manufacturers, and improvements in procurement channels for the DRAM required by enterprise SSDs could allow tight supply-demand conditions to persist through CY28. Goldman Sachs also emphasizes that NAND's cyclical nature has not disappeared: compared with the HDD and DRAM markets, which are dominated primarily by two to three companies, NAND still has more participants. However, the report expects profit levels in this cycle to exceed those of previous cycles and remain elevated for longer, with earnings over the next two to three years potentially exceeding market expectations. Based on the combined rationale of earnings sustainability, potential shareholder returns, and valuation rerating, Goldman Sachs reiterates its Buy rating and 12-month target price of ¥116,000. The target price is derived from the relationship between ROE and P/B and the average ROE forecast for FY3/27—FY3/28, implying a forecast FY3/28 P/E of 7.0x and P/B of 5.7x.

Analysis framework

Goldman Sachs first assesses the impact of the new fab building and medium-term capex on cash flow, then updates its shipment volume, cost, pricing, profit, and EPS forecasts based on research conducted after the first quarter. The report subsequently uses cash exceeding the equivalent of four months of sales as the surplus-cash threshold to estimate potential buybacks and shareholder returns, while drawing on changes in capital returns and valuations among Japanese shipping stocks after a cyclical peak. Finally, the report determines the target price by combining global memory peer valuations, the NAND supply-demand outlook, and the ROE-P/B relationship.

Methodology notes

• Valuation methodologyPB valuation

  ROE-P/B relationship-based valuation

  The report uses the relationship between Kioxia's ROE and price-to-book ratio, together with the average ROE forecast for FY3/27—FY3/28, to derive a 12-month target price of ¥116,000; this implies an FY3/28 P/E of 7.0x and P/B of 5.7x.

• Valuation methodologyPE/PEG valuation

  Cyclical and peer P/E comparison

  The report compares Kioxia's forecast FY3/28 P/E of slightly above 3x with the low- to mid-single-digit multiples of global memory peers and the 2—3x P/E ratios once recorded by Japanese shipping stocks, illustrating that cyclical concerns are already reflected in the valuation.

• Company fundamentals and financial frameworkFree cash flow analysis

  Surplus cash and shareholder return estimation

  Goldman Sachs defines cash exceeding the equivalent of four months of sales as surplus cash and assumes that, beginning in FY3/28, most surplus free cash flow will be used for shareholder returns, primarily through buybacks, resulting in an estimated cumulative two-year shareholder return of nearly 20%.

• Industry/sector analysis frameworkSupply-demand framework

  NAND supply-demand and pricing trajectory analysis

  The report combines factors including enterprise SSD and AI demand, tight HDD supply, equipment export-control risks, near-term incremental supply, and memory manufacturers' prioritization of DRAM investment to conclude that NAND supply-demand conditions may remain tight and support price increases.

• Industry/sector analysis framework

  Cross-industry historical cycle analogy

  The report uses the experience of Japanese shipping stocks, which continued to drive share prices and valuations higher through sustained buybacks after the container shipping market peaked, to explain how cyclical memory stocks could be repriced through stable shareholder returns.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Kioxia Holdings (285A.T)

  The report believes tight NAND supply-demand conditions, cost competitiveness, progress in data center products, and potentially expanded buybacks collectively support earnings and valuation upside.

  Strengths

  The world's third-largest NAND flash memory manufacturer; cost competitiveness supports higher margins; gradual progress in enterprise SSD and data center products; expected capacity to generate strong free cash flow.

  Weaknesses

  The NAND market has more participants than the HDD and DRAM markets, and industry cyclicality remains unchanged; longer delivery cycles for enterprise SSDs have led to lower bit shipment forecasts.

  Comparison

  The forecast FY3/28 P/E is slightly above 3x, while global memory peers trade at low- to mid-single-digit multiples; the report also compares Kioxia with Japanese shipping stocks whose P/E ratios once fell to 2—3x at a cyclical peak before sustained shareholder returns drove a rerating.

  Risks

  A slowdown in AI investment, the rise of Chinese NAND manufacturers, margin pressure from fluctuations in costs or capacity utilization, a sharp appreciation of the yen against the US dollar, and slowing NAND demand in non-AI applications.

Key data

• Rating and target priceBuy; 12-month target price of ¥116,000Current price of ¥52,500, with the report indicating 121.0% upside
• FY3/28 forecast valuationP/E 3.2xDescribed in the main text as slightly above 3x; global memory peers trade at low- to mid-single-digit multiples
• Potential shareholder returnsCumulative shareholder return of nearly 20% over FY3/28—FY3/29Calculated based on the current market capitalization and expected to consist primarily of share buybacks
• Recent share buybacksApproximately ¥800bnThe report cites this as evidence that sustained shareholder returns could drive valuation expansion
• Medium-term investment plan in JapanApproximately ¥5tn through 2032Combined investment by Kioxia and SanDisk covering the six years from FY3/27—FY3/32
• Kitakami K3 fab building investmentApproximately ¥1.8tnOperations are expected to begin in FY3/30, with the actual timing dependent on market trends
• Three-year average capexApproximately ¥470bn annually over FY3/27—FY3/29The overall plan is unchanged, although some spending may be brought forward slightly into FY3/27 and FY3/28
• New operating profit forecastsFY3/27 ¥8,426.6bn; FY3/28 ¥11,381.4bn; FY3/29 ¥13,578.5bnPrevious forecasts were ¥8,286.5bn, ¥11,371.8bn, and ¥13,463.9bn, respectively
• New EPS forecastsFY3/27 ¥11,011.9; FY3/28 ¥16,476.9; FY3/29 ¥24,167.5Previous forecasts were ¥10,827.8, ¥14,918.7, and ¥17,669.1, respectively
• Relative to consensusApproximately 10%—15% higherRefers to Goldman Sachs' FY3/28—FY3/29 operating profit forecasts relative to Bloomberg consensus
• NAND pricing trajectoryAt least through mid-CY27Quarterly prices are expected to continue rising moderately, with no material change to average selling price assumptions
• Target price implied valuationFY3/28 P/E 7.0x, P/B 5.7xThe target price is based on the ROE-P/B relationship and the average ROE forecast for FY3/27—FY3/28

Impact & implications

The report believes the new capex plan primarily changes the timing of certain investments without increasing the average annual capex framework for FY3/27—FY3/29, and therefore does not necessarily constrain the scope for shareholder returns beginning in FY3/28. If NAND supply-demand conditions remain tight, profits stay elevated, and the company sustains buybacks, the market discount reflecting concerns about the unsustainability of earnings at the cyclical peak could narrow, allowing the valuation to move from a forecast FY3/28 P/E of slightly above 3x toward the 7.0x implied by the target price.

Risks

• AI investment, primarily by hyperscale cloud service providers, may slow.
• The rise of Chinese NAND manufacturers may intensify competition.
• Rising costs or fluctuations in capacity utilization may lead to lower margins.
• A sharp appreciation of the yen against the US dollar may weigh on earnings.
• NAND demand in non-AI applications may slow.

What to watch

• Monitor the specific production start date of the K3 fab building and how management adjusts the construction and operating schedule according to market trends.
• Monitor whether FY3/27 and FY3/28 capex is brought forward and whether the three-year plan of approximately ¥470bn in average annual spending remains unchanged.
• Monitor whether NAND prices continue rising quarter over quarter until at least mid-CY27 and whether tight supply-demand conditions persist through CY28.
• Monitor whether surplus cash exceeding the equivalent of four months of sales is used as expected for shareholder returns, primarily through share buybacks, beginning in FY3/28.