Morgan Stanley Roadshow: Murata's AI Server MLCC Sales Expected to Grow 85%-90% in F3/27, Maintains Underweight on Ibiden
Claire Weston
Morgan Stanley's roadshow verdict: AI servers are driving MLCC spec upgrades, not shortage-driven price hikes. Murata's AI-related MLCC sales are set to grow 85%-90% in F3/27, and the value gap between companies that can mass-produce high-value-add products and those riding cyclical pricing will keep widening.
What are AI servers actually pulling — shortages and price hikes, or spec upgrades?
Morgan Stanley completed 31 one-on-one institutional meetings across Singapore and Hong Kong. The core conclusion: AI servers are driving component spec upgrades, not simple shortage-driven price hikes.
This means → companies that can reliably mass-produce high-value-add products capture a structural premium, while those riding cyclical pricing only catch a single wave.
In plain terms = among MLCC makers — MLCC being multi-layer ceramic capacitors, the tiny components that store charge on circuit boards — whoever can pack more capacitance into a smaller body is the real winner.
Where does Murata's 85%-90% growth come from — is the same capacitor just getting pricier?
Morgan Stanley forecasts Murata's AI / data-center MLCC sales to grow 85%-90% year-on-year in F3/27: volume up roughly 40%, ASP up roughly 50%.
The report stresses: the ASP increase is not the same product at a higher price — it reflects a rising mix of higher-priced, higher-value-add products.
This reflects a structural dynamic: each GPU generation roughly doubles total capacitance demand on the accelerator card, yet board space does not expand proportionally — smaller, denser MLCCs become a hard requirement.
Murata achieves this through finer barium-titanate particles, more dielectric layers, and narrower external electrodes. In plain terms = they stack more charge-storing layers into a component the size of a grain of rice.
Where does Murata's AI revenue share stand — how much room is left?
AI / data-center MLCC as a share of Murata's total MLCC sales: roughly 10%-15% in F3/26, rising to 20%-25% in F3/27. Morgan Stanley expects a similar pace thereafter.
This means → AI-related revenue is shifting from a "supplement" to the core growth engine of Murata's MLCC business — yet at under a quarter of the total, the runway is still opening up.
The report maintains Murata at Overweight.
Who can compete with Murata for this business?
Morgan Stanley's 2025 MLCC market-share estimates: Murata 40.8%, Samsung Electro-Mechanics 22.5%, Taiyo Yuden 11.3%, TDK 6.9%, Yageo 5.4%, others 13.0%.
In AI-server high-value-add MLCCs, the competitive ranking is: Murata first, Samsung Electro-Mechanics second, Taiyo Yuden third.
The key bottleneck: for high-demand specs — 1608-size 100μF, 1005-size 47μF, 0603-size 10μF — Morgan Stanley believes only Murata can reliably mass-produce the full range today.
Taiyo Yuden benefits from AI demand too, but froze capex in H2 2024; F3/26 MLCC capacity growth is estimated at only about 5%. This means → even with orders in hand, it may not have the capacity to fill them. Morgan Stanley estimates its AI / data-center MLCC sales grow 82%-83% year-on-year in F3/27, reaching roughly 15% of the mix.
Is commodity MLCC pricing power a good thing or a bad thing?
Morgan Stanley takes a restrained view on commodity MLCC price hikes: Murata does not actively push pricing on commodity products.
This means → Murata's logic is that short-term price hikes look good on paper but lower the entry barrier, making it easier for Chinese, Korean, and Taiwanese competitors to move in — costing market share over the medium to long term.
In plain terms = raising prices on commodity products is like broadcasting to rivals that "this business has great margins — come take a slice." Murata would rather channel resources into R&D and win on technology barriers, not price barriers.
The report notes that some competitors maximize short-term profit through broad-based price hikes yet do not invest enough in R&D and manufacturing improvements — a strategy that cannot sustain medium- to long-term share gains.
What is wrong with Ibiden — earnings are growing, so why Underweight?
On ABF substrates — high-end packaging substrates that connect chips to circuit boards — Morgan Stanley expects Ibiden to benefit from NVIDIA Rubin ramp: volume shipments began in Q4 F3/26, and Rubin-related ABF substrate sales in Q1 F3/27 will surpass Blackwell-related sales.
However, EMIB-T — an embedded bridge packaging technology — is unlikely to contribute meaningfully to profits until after F3/29, and its margins may not reach the level of NVIDIA's existing high-value-add products.
This means → earnings are growing, but the current share price already bakes in aggressive assumptions. If the pace of earnings growth cannot match what the valuation implies, the stock faces downside pressure.
The report maintains Ibiden at Underweight.
The perception gap from the roadshow: what are investors asking, and what does Morgan Stanley think they are missing?
BULL
High-value-add growth underappreciated
Investors focused on commodity pricing upside, largely overlooking the earnings expansion from high-value-add product growth.
Murata re-rating potential
Morgan Stanley sees this perception gap as a key driver of Murata's potential re-rating.
BEAR
Ibiden expectations too high
Ibiden's core risk is whether earnings growth can match the aggressive assumptions embedded in the current valuation.
EMIB-T contribution delayed
Meaningful profit contribution likely only after F3/29 — limited near-term catalysts.
In plain terms = the market is fixated on the short-term 'price hikes' story, but Morgan Stanley argues the real value divergence lies in who can keep making products others cannot — Murata is undervalued, Ibiden is overvalued.
Content is for reference only, not financial advice.