1 of 32

Advanced Packaging for AI/HPC

September 10, 2026

Charles Woychik

cwoychik@nhanced-semi.com

630-561-6813

2 of 32

Recent Announcements

EPS 2026 Binghamton University

3 of 32

Photonics Advantage

EPS 2026 Binghamton University

3

As the demand for faster and more efficient data processing grows,

photonics will be a critical component of future technological advancements

© Quantum Computing Inc.

Low energy consumption

High-bandwidth & fast processing

Precision & sensing

Miniaturization & scalability

Data over distances

All At Room Temperature

Photonics enables the mass production of quantum hardware by adapting standard semiconductor foundry processes

4 of 32

EPS 2026 Binghamton University

Founded in 2016 with a seasoned core team

U.S. Owned & Headquartered in Batavia, IL

AP Technology Leadership in Hybrid Bonding Processes

Extensive Experience with 2.5D & 3D Heterogeneous Integration

90+ Employees with Industry Recognized AP Experts on Staff

“Engineer of the Year”

Dr. Sangki Hong

“Small Business Supplier of the Year”

Recognized for our role in the building of 2.5D Interposer Assemblies

Article: “Compute Cambrian Explosion”

“Fabs and Labs Department”

Our Story & Recognition

5 of 32

NHanced Background

  • We target specialty niche markets with Advanced Packaging,

Additive Manufacturing, and specialty BEoL:

    • Photonics
    • Fluidics
    • True Heterogenous Integration
    • RadHard
    • Quantum devices
    • FPGAs
    • Memory including ReRAM, MRAM, FeRAM, EEPROM, DRAM, SRAM
    • AI/ML/Exascale
    • RFID
    • Medical
    • Down-hole to deep-cryo
  • Mixed Commercial and A&D customers
    • Development and production

EPS 2026 Binghamton University

Batavia IL, Headquarters

Design engineering,

Test, board level work, and back office.

Morrisville NC, Fab 1

Prototyping, pilot line, and LV manufacturing

~7k sq. ft. Fab area

>100 tools

Manual tools – high customization

Odon IN, Fab 2

“HVM” manufacturing

~7k sq. ft. Fab area

Automated/Semiautomated tools – high throughput, CPK monitor controlled

6 of 32

Specialized Materials:

  • Interposers: silicon, glass, and fused silica
    • Optical and electrical interconnect
  • Advanced substrates: silicon, GaN, GaAs, InP, GaSb, LiNbO3, BTO, SiC, TFLN, etc.

​

AP Assembly:

  • Flip-chip
  • Solder ball drop
  • Co-packaged optics
  • Interposer and substrate

​

​

​

Advanced Packaging:

  • HDI edge packaging
  • Chiplet integration
  • TSV insertion
  • Covalent oxide bonding
  • Hybrid copper and nickel bonding
    • True heterogeneous integration
  • Transfer printing assembly
  • Wafer reconstitution
  • Die-to-die, die-to-wafer, and wafer-to-wafer 2.5D and 3D integration

​

​

​

EPS 2026 Binghamton University

NHanced Standard Capabilities

ISO 9001 AP Manufacturing of:

Photonics

Fluidics

True Heterogenous Integration

RadHard

Quantum devices

FPGAs

​

​

Memory including ReRAM, MRAM,

FeRAM, EEPROM, DRAM, SRAM

AI/ML/Exascale

RFID

Medical

Down-hole to deep-cryo

​

7 of 32

EPS 2026 Binghamton University

8 of 32

Key Tools and Capabilities

  • Die2Wafer & Wafer2Wafer Bonding
    • BESI Next Gen
      • Better than 200nm accuracy
      • >2k die per hour
      • >three 9s going to five 9s
    • Laurie
  • Bosch Etch
    • PlasmaTherm
  • Oxide/Nitride
    • Trion
      • Low temperature – down to 70C oxides
      • Stress balanced
    • P5000
  • Pd/Au Plating
    • Quantum device enablement
    • UBM
  • Large TSV Fill
    • 50x300um, 25x200um

​

EPS 2026 Binghamton University

  • Oxide Etch
    • AMAT P5000s
  • PVD
    • Endura 5500
  • Metrology
    • Next gen ultrasound
    • X-ray
    • CDSEM
  • Production Plater
    • Raider
  • Solder Plating
    • Lead Free
  • Solder Ball Drop
    • Lead Free
    • Leaded
    • In
    • Bi/X
  • RIE Metal Etch
    • Oxford

​

  • I-Line Litho
    • NXQ8000
      • Full wafer reticle – stitchless interposers to full wafer size
      • Near I-line stepper performance: 1um L/S
      • Through wafer IR alignment
    • Ultra-Tech Saturns
  • ALD
    • High aspect ratio TSV barrier and seed
    • High k thin films
  • Extensive CMP
    • Oxide, Silicon, Barrier, Cu, Ni, Ru
  • Wafer Grinding

9 of 32

NHanced Small/Medium Scale � – High Touch Manufacturing

EPS 2026 Binghamton University

    • 2.5/3D Advanced Packaging
    • Si Sensors
    • Commercial Development
    • Mil-Aero Development and Manufacturing
        • Full BEoL
          • Niche BEoL Processes
          • Cu, Ni, Al, +
        • Split-Fab
        • High k-caps
        • ReRAM
        • 3/5 and LiNbO3 integration
      • Thick Wafer & Thin Wafer Processing
    • ISO9001

​

    • Target market is customers needing

1 to 5000 wafers per year

10 of 32

NHanced High Volume Manufacturing� – Customized At-Scale Manufacturing

EPS 2026 Binghamton University

    • 2.5/3D Advanced Packaging
      • Hybrid Bonding
      • Customize Metal Stack
    • Interposers
      • Silicon
      • Glass
    • Target market is customers needing

500 to 50,000 wafers per year

11 of 32

NHanced Market Messaging

EPS 2026 Binghamton University

12 of 32

It’s All About The Wire

  • Wire length controls the delay
    • Span of control
  • Accounts for majority of power usage
    • Memory fetch is an extreme example

By the mid 90’s wire was already becoming the dominant delay limiting performance. The GaAs FPGA demonstrated this perfectly. The transistor performance improvement of GaAs didn’t matter for the FPGA. It was the wire.

​

EPS 2026 Binghamton University

3DInCities

13 of 32

Effective End of Moore’s Law

Moore’s Law was first and foremost a statement about economics. We could shrink transistors and build more of them for about the same cost.

    • This has been the basic premise of the semiconductor industry for 50 years and was true up until the last few years.
    • Today we can indeed shrink transistors further, but the cost per transistor no longer declines.

We can get something a little more compact

      • Perhaps a little less power
      • But we pay more for these features now.

EPS 2026 Binghamton University

Intel

210 Scaling in 40 years

2µm → 2nm

One can see the atoms with this 6nm device. It is obvious we can’t scale much more.

Maybe to 1nm?

14 of 32

Moore’s Law Is About Cost

EPS 2026 Binghamton University

Chip Design and Manufacturing Cost under Different Process Nodes: Data Source from IBS

EETimes

Single pattern 193i lithography reaches its limit.

Smaller nodes required double patterning and/or EUV.

Lithography already the single most expensive step in semiconductor manufacturing doubles or triples in cost after the 28nm node – effectively ending Moore’s Law.

Runaway Development Costs

Performance Limited by Wire

15 of 32

Foundry 1.0: The Industry Today

  • Current semiconductor business has been focused on driving smaller transistors.
  • This is a safe path with 50 years of historic success!
  • But…
    • High development cost
    • High capital cost
    • Long development times
    • Expensive design tools
    • High risk
  • It is the Twilight of Moore’s Law

EPS 2026 Binghamton University

TSMC

16 of 32

What Does This Mean?

The semiconductor industry is about to undergo a sea change.

  • New ways of accomplishing Moore’s law economics and performance are needed.
    • The industry is now looking to use advanced packaging to drive future semiconductors.
      • Better Cost
      • Better Performance
      • Better SWaP

EPS 2026 Binghamton University

The revolution in semiconductors is reminiscent of the revolution in locomotives. The locomotive builders reached a tipping point driven by cost. After which, diesels which had been around for 20 years, almost over night replaced steam locomotives. The steam engine incremental improvements no longer were ample to sustain the industry.

The economics drove the change.

17 of 32

Foundry 2.0 –

A Finishing Foundry that takes the standardized building blocks from traditional semiconductor manufacturers and uses advanced packaging and additive manufacturing to create highly customized components with superior performance targeting small and medium sized markets.

EPS 2026 Binghamton University

A New Semiconductor Industry Paradigm

18 of 32

Foundry 2.0

  • System solution focus
    • Best of class components
  • Relies on Advanced Packaging (AP) and Chiplets
    • Heterogenous integration
    • Photonics
    • MEMS
    • RF
  • Advantages
    • Low development cost
    • Low capital cost
    • Short development times
    • Inexpensive design tools
    • Low risk

​

EPS 2026 Binghamton University

Stanford University

3DinCites

Philips

EETimes/Intel

Honeywell

19 of 32

AP Drivers: AI, Quantum, & Integrated Sensors

  • Advanced Packaging I/O
    • Tb/s, <<100fJ/b
      • SiP 500ff I/O Load
      • 2.5D 25ff I/O Load
      • 3D 3ff I/O Load
  • AI Musts:
    • Local Memory
    • >100TB/s

​

EPS 2026 Binghamton University

Intel

Tyndall National Institute Carroll et al

Nature 29 January 2021 Shastri

Honeywell

20 of 32

AP Drivers: Internet Of Things

EPS 2026 Binghamton University

ImpactLab.net

The IoT requires sensors, processors, RF, energy harvesting, and often special materials to meet the industry needs.

21 of 32

More Than Moore

EPS 2026 Binghamton University

GEORGIA TECH PRC

These new technologies and IoT are requiring introduction of new materials and processes that are often incompatible with CMOS. Advanced Packaging provides a mechanism for the tightly coupled integration of these new elements into true complete systems in a package with often far superior performance and lower power.

22 of 32

Hybrid Bonding – A Keystone Technology

EPS 2026 Binghamton University

The Enabler for Next Generation Advanced Packaging.

DBI® Advantages

    • Higher Bandwidth
    • Reduced Latency
    • Increased Efficiency
    • Higher I/O Density
    • Scalable (<1µm Pitch)
    • Low Temperature Bonding

1st bonding: F-to-F

1st bonding: F-to-F

1st bonding: F-to-F

1st bonding: F-to-F

2nd bonding : B-to-B

2nd bonding : B-to-B

3rd bonding : B-to-B

23 of 32

DBI®: Low Temperature Hybrid Bonding Process

EPS 2026 Binghamton University

Jihoon Seo / Materials Research Society

24 of 32

EPS 2026 Binghamton University

0.1 micron alignment

1.9 µm DBI® pitch, 300℃

1.6 µm DBI® pitch, 300℃

Scalable To < 1um Pitch

0.8um Pitch Demonstrated

Cu/SiO DBI® Hybrid Bonding

Hybrid Bonding Interconnect Pitch Scaling

    • WtoW 3sigma < ±1um misalign performance
    • DtoW 3sigma < ±100nm misalign performance
    • Production Minimum pitch = 2um

P=3.2um

0.3um

25 of 32

Driving Die-to-Wafer Progress

  • The BESI Datacon 8800 CHAMEO ultra plus AC D2W bonding system that can reach >200nm of accuracy for each bond.
  • To reach this level of accuracy the tool has demanding requirements that need to be fulfilled like:
    • Advanced tooling for the tool
      • Time and Cost
      • Plasma singulation needed
    • Specific types of key patterning and placement
      • Clear field
      • DRC waivers
    • Fiducial material design
      • Contrast
    • Film frame picking

EPS 2026 Binghamton University

26 of 32

AP Cornerstones: Embedded 2.5/3D ─ Wafer Reconstitution

EPS 2026 Binghamton University

ASE

HRL

DECA Technologies

Wafer reconstruction and adaptative substrates provide additional flexibility and next level packaging alternatives.

27 of 32

AP Cornerstones: Chiplets

  • Provides:
    • Best of Class Everything
    • Easy retargeting
    • Lower risk
    • IP reuse
    • Lower cost

​

EPS 2026 Binghamton University

Intel

Intel

AMD

nVidia

Z-Glue

28 of 32

AP Cornerstones: Interposers

EPS 2026 Binghamton University

  • Up to full wafer scale interposers
  • HP 2um thick wiring is standard
  • 4um, 6um, 10um, layer options
  • Embedded decoupling caps and 12 wiring layers coming
  • Optical waveguides

Mosaic Microsystems

29 of 32

AP Extensions: Cooling, Microfluidics and Diamonds

EPS 2026 Binghamton University

Chip Scale Cooling

For Ultra-Dense AI Electronics

30 of 32

AP Extensions: Micro-Connections and 3D Structures

EPS 2026 Binghamton University

(top left) Quilt Packaging concept cross-section illustration & image of a post-reflowed QP CMOS quilt; (bottom left) SEM image of quilted chip-to-chip seam of >10 micron width; (top right) QP-enabled miniature curved array demonstration article; (bottom right) profile view of QP-enabled miniature curved array.

IIC

Bacteria-World

techunwrapped.com

MEMS + Precision Interconnects

31 of 32

Why Are We Headed To Low/Medium Volume Manufacturing?

  • Transistor cost is increasing
    • To reduce cost, we need to sell the consumer fewer transistors
    • Most semiconductor devices have the kitchen sink
      • NRE is high so we can’t target markets
  • Chiplets offer reuse – effectively $0 NRE
  • Advanced packaging NREs run ~100x less cost than lead edge node NREs
  • Chiplet to chiplet power and delay is competitive with on die
  • Chiplets can target markets – use only necessary transistors

​

→ Low NRE → Niche Market → Fewer Transistors → Lower Cost Composite Device

​

EPS 2026 Binghamton University

Moore’s Law for the Next Decade

32 of 32

A Change In Perspective Foundry 2.0

  • A new focus on next generation semiconductors created by
    • Advanced Packaging
    • Additive Semiconductor Manufacturing
    • True heterogenous integration
    • Interconnect focused – all BEoL additive
      • Better ROI
      • Lower development costs
      • Lower CAPEX
    • Leveraging existing foundries
      • Split Fab
    • More Than Moore Technologies
      • IP Centric
      • IP driven value – not cost of capital

​

System Level Moore’s Law

EPS 2026 Binghamton University

Split Rock Lens – StackExchange