Aug Concentration: Median Top-1 70.5%; CoWoS Dual Tip 75%/52% — Test Tools +31% vs Packaging +9.6%

Our mid-window concentration companion mapped the chip stack as a family of top-1 / top-3 shares. The Q3 concentration refresh then re-metered the tip after the CoWoS tracker: median top-1 eased 72% → 71%, HBM firmed to ~55%, and CoWoS supply tip slipped to ~76% while NVIDIA buyer share stayed ~55%. The August Mid-Year back-end cut changed the flow story — test equipment $15.3B (+31%) versus assembly & packaging tools only $6.7B (+9.6%) on a $165.9B Mid-Year path. This August 202608 concentration lens asks the distribution question again: did that growth asymmetry rewrite where the stack concentrates at the top — or only explain why the packaging gate still binds?

The dashboard above is built as a dual-tip + growth-concentration board. Toggle Dual tip, Layer ladder, Growth conc., and Geo / gates. The punchline is deliberately two-sided. Median top-1 eases 71% → 70.5% and stack HHI softens ~5,620 → ~5,550 — mild tip softening, not a regime change. CoWoS supply tip slips 76% → 75% while buyer tip softens 55% → 52% at 140k wpm. EUV remains 100% ASML; leading-edge foundry stays ~90% TSMC; HBM holds ~55% SK Hynix inside a closed three-player set. The Mid-Year print’s 21.4-point test-versus-packaging YoY gap is the growth-concentration input that keeps CoWoS binding even as supply ownership dilutes a little.

The August scoreboard: top-1, Δ vs Q3, and HHI

Read the table as a family of market shares with a Q3→Aug Δ column, not one semiconductor “concentration number.” Most of the ladder is structurally flat this vintage — EUV, foundry tip, HBM, EDA, wafers, and broad WFE do not move. The movers that matter for AI shipping risk are GPU (−1 pp tip) and CoWoS (−1 pp supply tip) — small, same-direction tip easing that still leaves five of eight layers at top-1 ≥ 70%.

Dual tip: supply ownership is not buyer concentration

Open Dual tip. The composed panel puts CoWoS capacity (bars, thousand wafers per month) against two share lines: TSMC supply tip and NVIDIA buyer tip. Across theme vintages the geometry is sticky: capacity rises toward 140k wpm, the supply tip drifts down from the research-era ~80% toward ~75%, and the buyer tip oscillates in a ~52–55% band rather than collapsing. The dashed gap line still reprints near ~20% on the Mid-Year cut.

That is why “more CoWoS capacity” is not the same sentence as “less concentration.” Supply-side tip softens a little as OSAT and rival advanced-packaging ramps add wafers; demand-side concentration at the tip of the queue softens only to ~52% — still a majority reservation narrative for a single buyer. The Q3→Aug dumbbells beside the dual path show most layers flat; GPU and CoWoS are the visible movers. Stage averages still put midstream near ~73% average top-1 and downstream at ~86%. For the capacity-versus-demand frame that feeds this lens, see the Q3 CoWoS tracker and the August Mid-Year update.

Layer ladder: saturation still happens before top-3

Switch to Layer ladder. Rank by top-1, top-3, or HHI; filter by stage. The companion top-k path still shows EUV, foundry, and GPUs near ceiling at top-1; HBM and CoWoS still climb into top-2/top-3; WFE never saturates. The Lorenz-style mass panel under the path remains well above an equal-share guide: the first few layers still carry most of the single-vendor risk mass in the stack.

Relative to the Q3 concentration companion, the qualitative ordering is unchanged. What changed is the meter: median top-1 ticks 71% → 70.5%, stack HHI softens ~5,620 → ~5,550, and CoWoS remains above the 70% top-1 bar even after another 1 pp supply-tip ease. Toggle Movers only on the Δ panel to hide the structural ceilings — EUV’s flat 100% and foundry’s flat ~90% are not “no news”; they are the reminder that Mid-Year equipment dollars and AI demand did not create a second EUV vendor or a second leading-edge foundry tip this vintage.

Growth concentration: the Mid-Year back-end punchline

Open Growth conc. The Mid-Year OEM segment bars are the August contribution. Toggle YoY % versus Level $B. DRAM tools +39% to $38.8B and test equipment +31% to $15.3B race ahead of the stack; assembly & packaging tools only +9.6% to $6.7B on a total equipment path of $165.9B (+23.2%). The asymmetry panel states the gap in one view: 31.0 − 9.6 = 21.4 points.

That is growth concentration, not vendor HHI inside each segment. Upstream dollars fund the stack; slow packaging-tool growth helps explain why CoWoS still gates the ship even when broad WFE top-3 is only ~56%. The HBM donut and foundry dual bars underneath restate the midstream structure: SK Hynix ~55%, Samsung ~33%, Micron ~12% — top-3 still equals the market — and TSMC ~90% advanced-node versus ~62% overall. A held HBM tip inside a closed three-player set is still more concentration risk for GPU ship schedules than a diversified memory census would imply, because qualification cycles and HBM3E/HBM4 ramps do not create a fourth supplier on a slide.

Geo / gates: install geography plus the packaging dual tip

Open Geo / gates. SEMI’s 2025 WWSEMS billings still put China + Taiwan + Korea at ~79% of global equipment dollars — a geographic top-3 that rivals vendor concentration inside several layers. China alone is ~36.5% of billings; Taiwan’s AI-install signature from the research map remains intact. Q1 2026 billings at $36.55B (+14%) are the pace check against the Mid-Year annual path, not a rewrite of install geography.

The CoWoS close-up restates the dual tip without the capacity bars: supply 75%, buyer 52%, gap ~20%. That geometry is why desks that quote only “TSMC CoWoS share” or only “NVIDIA reservation share” miss half the concentration object. For the packaging-and-memory bottleneck frame that precedes these vintages, see GPU packaging & memory.

Why WFE still looks diversified next to EUV

The analytical error the mid-window and Q3 posts warned about remains the live risk: collapsing “semiconductor equipment” into one concentration statement. ASML’s EUV monopoly is still 100%. Applied / ASML / Lam as a ~56% top-3 in broad WFE is also still real. Quoting either without the other mis-states resilience. The August twist is that even diversified WFE growth does not unlock the packaging gate when assembly & packaging tool growth lags by more than 20 points. Export controls, tool lead times, and service networks still create friction in diversified WFE — but the industrial-organisation object is different from a single-scanner monopoly. EDA sits between those poles: Synopsys + Cadence still clear roughly ~72%, and top-3 approaches ~88%.

Caveats

• Layer shares are public narrative bands and company-disclosure composites, not a single SEMI census of vendor revenue by AI-relevant slice. Treat advanced-foundry and GPU percentages as order-of-magnitude structure, not filing-precision microdata.
• Δ columns are editorial vintage comparisons against the Q3 concentration companionuseful for ranking what moved, not a formal market-share audit trail.
• HHI figures are derived from stated top buckets plus a residual; they are analytical meters for ranking layers, not antitrust exhibits.
• Tightness scores are editorial (1–10) carried from the theme’s bottleneck framework; they are not SEMI metrics.
• CoWoS dual tip mixes capacity ownership and buyer reservation narratives across theme vintagesuseful for concentration storytelling, not a substitute for TSMC’s full packaging P&L or NVIDIA’s disclosed purchase commitments.
• Segment YoY figures are Mid-Year OEM-forecast growth, not vendor HHI inside each equipment segment. Regional equipment shares are install geography (where tools ship), not where corporate parents are headquartered.

What to watch next

Watch whether CoWoS supply tip keeps easing as OSAT and Samsung advanced-packaging capacity scales — or whether TSMC’s AI-relevant share re-firms when demand re-accelerates faster than wafer-out. Watch the buyer tip: Mid-Year softens it to ~52%, but capacity growth without durable buyer diversification still leaves a majority reservation narrative as the demand-side concentration print. Watch HBM: a true fourth supplier would change the oligopoly math; incremental Micron share inside a three-player set would not. Watch the Mid-Year back-end asymmetry — if assembly & packaging tool growth stays stuck near +10% while AI wafer demand compounds, the packaging gate remains the binding midstream object even when front-end WFE looks “diversified.” And keep EUV as the upstream ceiling — High-NA does not create a second EUV vendor.

Primary sources: SEMI Mid-Year OEM Forecast (July 2026) and Q1 2026 billings; TrendForce / company disclosures for HBM and CoWoS reservation bands; theme posts Q3 concentration, mid-window concentration, August Mid-Year back-end cut, Q3 CoWoS tracker, research, and packaging & memory.

Bottom line: The August Mid-Year concentration lens does not rewrite the ladder of extremes — it re-meters the tip and names the growth gap that keeps midstream binding. Median top-1 eases 71% → 70.5%; CoWoS dual tip prints ~75% supply / ~52% buyer at 140k wpm; test tools +31% versus packaging tools +9.6% leave a 21.4-pt asymmetry. Upstream/downstream bottlenecks are still where high share and high tightness coincide — and quoting one layer’s Δ, or one equipment segment’s YoY, as “the” semiconductor deconcentration story is how desks get the risk wrong.