Bspdn Weigh Three Commercial Implementations Side by Side
# Backside Power Delivery: Step 18 — Weigh the Three Commercial Implementations Side by Side
## 1. One Architecture, Three Executions
Steps 15 through 17 introduced Intel PowerVia, TSMC Super Power Rail, and Samsung's BSPDN
approach individually. All three implement the same underlying architecture this series has
traced throughout -- wafer flip and thinning, nano-TSV formation, and a buried-rail interface
to the transistor layer -- but differ in naming emphasis, public introduction timing, and the
specific transistor generation each foundry has paired its implementation with.
## 2. Differences in Emphasis, Not Just Name
Intel's public positioning for PowerVia emphasizes the backside delivery network reaching into
the frontside signal stack near the transistor. TSMC's naming for Super Power Rail emphasizes
the buried-rail structure specifically. Samsung's disclosures describe its approach in terms of
pairing with its own gate-all-around transistor family. These differences in emphasis reflect
each foundry's own process choices and public communication strategy more than a difference in
the underlying physics -- all three must solve the same wafer-thinning precision, nano-TSV
resistance, and buried-rail contact-junction challenges this series described in Steps 6, 7,
and 9.
## 3. Differences in Stated Timing
Each foundry has published its own introduction timeline, generally in the order Intel, then
TSMC, then Samsung, per Step 17 -- though public roadmap timing is subject to change and should
not be read as a technical ranking of implementation quality.
## 4. A Shared Set of Tradeoffs, Independently Managed
What all three share, regardless of naming or timing, is the full set of tradeoffs this series
has already quantified: IR-droop reduction on the order of five to ten times (Step 10),
standard-cell area shrink of roughly fifteen to twenty percent (Step 11), added process cost
and cycle time (Step 13), and wafer-bonding yield risk unique to backside processing (Step 12).
Each foundry manages that same set of tradeoffs independently, with its own equipment, process
recipes, and yield-learning curve.