Foundry allocation and lead-time reference
Silicon Analysts tracks booking-to-delivery lead times and allocation status across semiconductor foundry capacity: leading-edge logic (TSMC N2, N3, N5, N7; Samsung 2nm and 3nm; Intel 18A; SMIC 7nm; Rapidus 2nm), mature and specialty nodes (UMC, GlobalFoundries), advanced packaging (CoWoS, SoIC, WMCM) and HBM supply from SK Hynix, Samsung and Micron. Lead times are for new capacity orders rather than existing long-term agreements. 7 of 13 records carry a quantified lead-time band; the remainder are tracked qualitatively because no defensible public number exists. A quarterly time series covers CoWoS-S, CoWoS-L, HBM3E and TSMC N3 from Q1 2022, capturing the 2023 shortage peak and the subsequent capacity relief.
Foundry Allocation & Lead Times
CoWoS (S + L) runs 52 weeks booking-to-delivery and is fully booked. The longest quantified wait on a logic node is TSMC N2 / A16 at 78–156 wks. Lead times are for new orders, not existing LTAs — the number that decides whether a design can ship next year.
Baseline: April 2026 ·
How current is this? The allocation table is a curated baseline last rebuilt in April 2026. Weekly signal tracking is running; no newer confirmation is published yet.
Lead-time history (quarterly, weeks)
CoWoS-S, CoWoS-L, HBM3E and TSMC N3, Q1 2022 through Q1 2026 — the series runs to the most recent quarter with a defensible public reading, not to today. Annotations mark the 2023 demand surge, the Q4 2023 shortage peak and the capacity relief that followed.
Pull this series into a spreadsheet: /api/v1/market-data/component-lead-times?format=csv · Excel & Google Sheets setup →
Current allocation by node
| Foundry | Node | Status | Lead time |
|---|---|---|---|
| TSMC | N2 / A16 | Fully booked | 78–156 wks |
| TSMC | N3E / N3P / N3B | Fully booked | 104–156 wks |
| TSMC | N5 / N4P / N4NP | Constrained | 22–32 wks |
| TSMC | N7 / N6 | Available | 8–16 wks |
| TSMC | N16/12 / N28/40 | Available | 4–12 wks |
| Samsung | SF2 / SF2P (2nm) | Constrained | Not quantified |
| Samsung | 3GAE / 3GAP (3nm) | Available | Not quantified |
| Samsung | 4LPP (4nm) | Constrained | Not quantified |
| Intel | Intel 18A | Early access | Not quantified |
| GlobalFoundries | 12nm-22nm | Available | 10–16 wks |
| SMIC | N+2 (7nm) | Severely constrained | Not quantified |
| UMC | 14nm-28nm | Available | 8–14 wks |
| Rapidus | 2nm GAA | Early access | Not quantified |
“Not quantified” means no defensible public lead-time number exists for that node — status and capacity are still tracked. Lead times cover new capacity orders; existing long-term agreements are priced and scheduled separately.
This table as CSV: /api/v1/allocation?format=csv
Advanced packaging
CoWoS (S + L)
TSMC + OSATs
CoWoS-L dominates new expansion (required for dual-die Blackwell/Rubin). NVIDIA holds >70% of CoWoS-L specifically. OSAT outsourcing: ~240-270K wafers/yr. Lead times >50 weeks for new orders.
Customer Allocation
2026 target: 130K WPM
SoIC (3D hybrid bonding)
TSMC
TSMC VP confirmed >100% CAGR 2022-2026. 4 major companies currently using SoIC. AP7 Phase 1 targets SoIC mass production 2027.
Customer Allocation
2026 target: 20K WPM
WMCM (Wafer-Level MCM)
TSMC
Hybrid of InFO and CoWoS. Ramping to ~60K WPM by end 2026, >120K WPM by 2027. Dedicated primarily to the lead smartphone platform (public reports).
Customer Allocation
2026 target: 60K WPM
CoWoS is the binding constraint on AI accelerator supply. SoIC and WMCM are tracked by status and capacity ramp only — neither has a public booking-to-delivery figure we can stand behind.
HBM supply
SK Hynix
53% market share
HBM3E: Fully allocated through 2026+. 12-Hi in full mass production.
HBM4: Mass production started Feb 2026 at M16 Icheon + M15X Cheongju. Using TSMC 12nm for base die.
Dominant position. M15X started 4 months ahead of schedule. FY2025 revenue KRW 97.1T (+47% YoY), 58% operating margin in Q4. CapEx $20.5B for 2026.
HBM3E $/stack
$330
HBM4 $/stack
$520
NVIDIA HBM4 allocation: ~55%
Key Customers
NVIDIA (primary) · Google · AMD
Samsung
35% market share
HBM3E: In mass production. NVIDIA qualified Sep 2025 after 18 months of failed attempts.
HBM4: World's first commercial HBM4 shipped Feb 12, 2026. Vertically integrated: 1c DRAM + 4nm foundry base die.
Dramatic recovery from 17% share in Q2 2025 to 35%. Turnkey advantage: memory + foundry + packaging under one roof. HBM sales expected to more than triple in 2026.
HBM3E $/stack
$310
HBM4 $/stack
$500
NVIDIA HBM4 allocation: ~25%
Key Customers
Google (>60% of Google HBM) · Broadcom · NVIDIA · AMD
Micron
11% market share
HBM3E: Full production. 12-Hi yields surpassed 8-Hi. 30% lower power than competitors.
HBM4: Volume production started Q1 FY2026. 16-Hi samples (48GB) shipped. Meeting only 50-66% of customer demand.
Record FQ2 2026: revenue $23.86B (nearly 3x YoY), 74.9% gross margin. Produces both DRAM and base dies in-house. CapEx raised to >$25B for FY2026.
HBM3E $/stack
$320
HBM4 $/stack
$510
NVIDIA HBM4 allocation: ~20%
Key Customers
NVIDIA · AMD · 4 paying customers total
HBM capacity is contracted through long-term agreements rather than booked at a quoted lead time, so these records carry allocation share and qualification status instead of a week count.
Recent allocation changes
- 2026-04-01
JASM Phase 2 upgraded from 6nm to 3nmTSMC N3
Taiwan government approved upgrading TSMC's Japan JASM P2 from original 6nm plan to 3nm, adding 15K WSPM of advanced-node capacity by 2028.
Source: Taipei Times
- 2026-04-01
Intel Fab 34 Apollo buyback for $14.2BIntel Intel 18A
Intel repurchasing Apollo's 49% stake in Leixlip fab at $3.2B premium, restoring 100% ownership. Funded by cash + $6.5B new debt.
Source: Tom's Hardware; Intel
- 2026-03-26
Samsung Taylor fab achieves EUV first lightSamsung Samsung 2nm
>90% construction complete. Risk production targeting H2 2026. Tesla $16.5B anchor tenant for AI5/AI6 chips.
Source: Tom's Hardware
- 2026-03-18
Micron reports record $23.86B quarterly revenueMicron HBM
Nearly 3x YoY growth with 74.9% gross margins. HBM4 volume shipments and 16-Hi samples shipped. CapEx raised to >$25B.
Source: Micron FQ2 2026 Earnings
- 2026-02-27
Rapidus raises $1.7B from 32 investorsRapidus Rapidus 2nm
Japanese government becomes largest shareholder at 11.5%. Canon confirmed as first customer. Mass production on track for H2 2027.
Source: Rapidus; Nikkei
- 2026-02-12
Samsung ships world's first commercial HBM4Samsung HBM4
Using vertically integrated 1c DRAM + 4nm foundry base die. Achieved "highest evaluation scores" in NVIDIA testing for speed and power efficiency.
Source: Samsung; KED Global
- 2026-02-01
SK Hynix begins HBM4 mass productionSK Hynix HBM4
Production at M16 Icheon and M15X Cheongju using 1b DRAM + TSMC 12nm base die. Mid-50% of NVIDIA HBM4 allocation.
Source: SK Hynix Q4 Earnings; Hankyung
- 2026-01-28
ASML records record Q4 bookingsASML Equipment
Nearly doubled consensus. Backlog through 2027. Memory orders surpassed logic for first time (56/44 split). 60+ High-NA systems ordered.
Source: ASML Q4 2025 Earnings
Latest entry 2026-04-01 — this log tracks structural allocation moves, not weekly signal readings.
Methodology
Lead times are booking-to-delivery for new capacity orders, compiled from foundry earnings calls, memory-vendor investor days, procurement reporting and supply-chain research. Existing long-term agreements are scheduled and priced separately and are not represented here.
The allocation table is a curated baseline. A weekly agent searches public sources for capacity signals, requires corroboration across independent sources before publishing, and stacks each weekly reading rather than overwriting the previous one — so a record shows how an estimate evolved, not just its latest value. Readings that are single-source or that move sharply are held for human review.
Where a node carries no lead-time band, that is a deliberate absence rather than a gap in collection: no public figure exists that survives corroboration. We would rather show “not quantified” than infer a number.
For full data provenance and terms of use, see Terms & Data Provenance.
Filter allocation by foundry, node or customer →
Allocation Explorer — the same records with per-node detail, customer concentration and capacity ramp targets.
What the constrained supply costs →
HBM Pricing — contract and blended $/GB by generation, dated and sourced.