Data centre power electronics for Japan need 100 V capacitors found by value, and a text search misses most of them: Walsin Technology (Walsin) lists 27 X7R parts at 2.2 microfarad and 100 V, 24 spelled 2200nF and only 3 spelled 2.2uF. OCCTO’s outlook, as summarised by Shulman Advisory, puts data centre peak demand near 6.6 GW by FY2035.
Venatech Limited, an authorised Walsin distributor in Hong Kong, wrote this for a power-supply or board design engineer at a Japanese data centre equipment maker who is crossing a 100 V bulk or bus capacitor. The market figures are sourced and dated below. The counts come from the Walsin MLCC export, file 00_ALL_MLCC_parts_combined.xlsx (24,026 rows, 23,601 distinct part numbers), and our ready-stock snapshot as at 23 September 2026. Every part number is real.
What does the OCCTO outlook say about Japanese data centre electricity demand?
OCCTO published its ten-year outlook for FY2026 to FY2035 on 21 January 2026; Shulman Advisory’s summary, dated 27 January 2026, reads about 6.6 GW of data centre peak demand and about 49 TWh of annual data centre consumption by FY2035.
The same summary gives about 7.6 GW of added peak demand for data centres and semiconductor fabs together, and notes that growth is back-loaded because projects were delayed and plans revised. These are the summary’s figures, not ours, and we have not read the OCCTO document itself. They tell a buyer that power-stage boards for racks are a long-running demand line, not what any named Japanese company buys.
Why does a 2.2 microfarad, 100 V X7R line return so many Walsin variants?
We parsed the Walsin MLCC export for X7R at 100 V and 1 microfarad or more and found 82 distinct part numbers across six case sizes, from 1206 to 2225; 27 of them sit at 2.2 microfarad.
They differ by series (general purpose, big case, soft termination, AEC-Q200 MT and ST, low profile), by tolerance and by thickness. The table puts one 1210 brief against six of them and leaves a column for your incumbent.
| Walsin part | Series, size, capacitance as spelled, tolerance, thickness | On shelf 23 Sep 2026 | Match / mismatch with a 2.2 microfarad, 100 V, X7R, 1210, plus or minus 10 percent line | Your incumbent |
|---|---|---|---|---|
1210B225K101 |
General Purpose Caps, 1210, 2200nF, plus or minus 10%, 100 V, T 2.50±0.50 | Yes | Match on size, value, tolerance, voltage and dielectric; on the shelf | |
1210B225M101 |
General Purpose Caps, 1210, 2200nF, plus or minus 20%, 100 V, T 2.50±0.50 | No | Mismatch: tolerance is plus or minus 20 percent, wider | |
1210B225J101 |
General Purpose Caps, 1210, 2200nF, plus or minus 5%, 100 V, T 2.50±0.50 | No | Mismatch: tolerance is plus or minus 5 percent, tighter | |
MT32B225K101 |
Automotive Caps Qualified to AEC-Q200, 1210, 2200nF, plus or minus 10%, 100 V, T 2.50±0.50 | No | Match on the electrical fields; mismatch in series: AEC-Q200 MT series | |
ST32B225K101 |
Automotive Soft term Caps Qualified to AEC-Q200, 1210, 2.2uF, plus or minus 10%, 100 V, T 2.50±0.50 | No | Match on the electrical fields; mismatch in series: soft termination ST, and the value is spelled 2.2uF | |
SH32B225K101 |
Soft Termination Capacitors, 1210, 2200nF, plus or minus 10%, 100 V, T 2.50±0.50 | No | Match on the electrical fields; mismatch in series: soft termination SH |
A match in the table is a match on export fields. Confirm against both datasheets and qualify in your own circuit. For the method, see how to cross-reference an MLCC.
Why does a search for 2.2uF find only 3 of 27 Walsin parts?
Because the export spells the same value two ways: 67 of the 82 parts at 1 microfarad or more are written in nanofarad (2200nF) and only 15 in microfarad (2.2uF). At 2.2 microfarad the split is 24 to 3; at 4.7 microfarad it is 8 spelled 4700nF and 2 spelled 4.7uF; at 10 microfarad all 7 are spelled in microfarad.
A filter or spreadsheet formula that matches the text 2.2uF returns 3 parts and hides 24. We have not seen this counted anywhere. The cross-reference tool reads the value, not the spelling, so paste the incumbent part and read the size and tolerance fields. One more oddity from the same count: 1812B225B101 prints a tolerance of plus or minus 0.1 pF against a microfarad value, which we treat as an export fault to confirm with the datasheet.
How much of the 100 V X7R rung is on the Hong Kong shelf?
Two of the 82 part numbers are on the shelf as at 23 September 2026, 1210B105K101 and 1210B225K101, both in 1210. Of the 82, 29 are in the big cases 1825, 2220 and 2225, which are mechanically a different design from a 1210 (see our note on DC bias and derating).
The ready stock list is a dated snapshot, never live-synced, and the MLCC range lists the families, so ask for a current check through the site. A 100 V part is often chosen for a 48 V bus; that is a design decision. Apply your own derating rule, because ceramic capacitance falls under bias, and we publish no suitability claim.
What will we not tell you about Japanese data centre power design?
We will not tell you which capacitor your converter needs, how much capacitance survives bias, or what any Japanese operator buys. We hold 28 Walsin datasheets, 451 pages, and none names a data centre.
We hold no ISO 9001 or IATF 16949 certificate and claim no office, entity or stock in Japan; our locations are Hong Kong, Shenzhen, Singapore and Jiangmen, with Chengdu in preparation. We give no legal, customs or conformity advice. Confirm with your compliance team. Qualification stays with you.
How can Venatech help a Japanese power-board team?
How Venatech helps starts with a free paste box. Our VenaMatch tool at /cross-reference/ decodes 18 manufacturers’ part numbers, covers 800,000+ crossable parts, takes up to 500 part numbers or an Excel or CSV file, and marks every field match or mismatch. Upload a bill of materials at /bom/ and check the ready stock snapshot dated 23 September 2026.
Worked example from our own export: a 2.2 microfarad, 100 V, X7R, 1210, plus or minus 10 percent brief has 27 variants at the value in all sizes, 1 exact shelf part, and 5 neighbours that differ in tolerance or series. Walsin is the only line we are authorised for; Stackpole and Viking are lines we supply. We work from Hong Kong, welcome small lots, and you request a quote through the site. We give no legal, customs or conformity advice; confirm with your compliance team, and qualification stays with you.
The short version
For a Japanese data centre power board, OCCTO’s outlook (as summarised by Shulman Advisory) points to about 6.6 GW of data centre peak by FY2035. In the Walsin export, the 100 V X7R rung at 1 microfarad or more is 82 parts, 67 spelled in nanofarad, and only 2 on our shelf.
Send us one 100 V bulk capacitor line from your power board with the tolerance and size you can accept, and we will return the Walsin parts with every difference stated. Does your part search read 2200nF as 2.2uF?
Sources (fetched 10 November 2026)
- Shulman Advisory, Data Center Market Update 2: OCCTO 10-year demand forecast, 27 January 2026, summarising the OCCTO outlook published 21 January 2026: https://shulman-advisory.com/data-center-market-update-2-occtos-10-year-demand-forecast/
- Walsin MLCC export 00_ALL_MLCC_parts_combined.xlsx and Venatech ready-stock workbook (.xlsx) as at 23 September 2026, re-measured 10 November 2026 (internal files).
Related reading
- How to cross-reference an MLCC, field by field
- MLCC DC bias and voltage derating
- AEC-Q200 MLCC: MT versus MG series
- Free cross-reference tool (VenaMatch)
- Ready stock, Hong Kong snapshot dated 23 September 2026
Last reviewed 10 Nov 2026.
Frequently asked questions
What does OCCTO expect for Japanese data centre demand?
Shulman Advisory’s summary of the OCCTO outlook published 21 January 2026 gives about 6.6 GW of data centre peak and about 49 TWh of consumption by FY2035. These are the summary’s figures; we have not read the OCCTO document.
How many Walsin X7R capacitors at 2.2 microfarad and 100 V are listed?
27 distinct part numbers in the export: 24 spelled 2200nF and 3 spelled 2.2uF. They differ in series, size, tolerance and thickness, and only 1210B225K101 matches a 1210, plus or minus 10 percent brief on every field.
Why does searching for 2.2uF miss most Walsin parts?
Because the export writes the same value as 2200nF on 24 of 27 parts. Search by value, not by text, and read the size and tolerance fields before you commit a footprint.
Is a 100 V capacitor the right choice for a 48 V bus?
That is your design decision. A 100 V rung gives voltage margin, but ceramic capacitance falls under DC bias. Apply your own derating rule and qualify in your own circuit. We make no suitability claim.
Does Venatech have an office or stock in Japan?
No. Our locations are Hong Kong, Shenzhen, Singapore and Jiangmen, with Chengdu in preparation. We serve Japanese buyers from Hong Kong, in small lots, with a quote requested through the site.
Can Venatech guarantee the stock shown?
No. The ready-stock list is a snapshot dated 23 September 2026 and is not live-synced. We state no price, lead time or quantity guarantee.