Chip employs TSMC CoWoS® technology to integrate SoC and HBM2
SAN JOSE, Calif., May 01, 2018 -- eSilicon, an independent provider of FinFET-class ASICs, custom IP, and advanced 2.5D packaging solutions, today announced that the deep learning ASIC that taped out last September has moved to production qualification.
The ASIC includes custom pseudo two-port memories designed by eSilicon, TSMC’s Chip on Wafer on Substrate (CoWoS) technology, 28G SerDes, and four second-generation high-bandwidth memory stacks (HBM2). eSilicon’s end-to-end 2.5D/HBM2 solution includes 2.5D ecosystem management, silicon-proven HBM2 PHY, ASIC physical design, 2.5D package design, manufacturing, assembly and test.
The CoWoS interposer is over 1,000 square mm and contains over 170,000 microbumps. The design has successfully passed test bring-up and is in final qualification. Four-high and eight-high HBM stack versions are in qualification. This design is in the industry vanguard of ASICs targeting deep learning applications.
The 2.5D/HBM2 single package implementation gives the ASIC many advantages:
- Orders of magnitude higher total bandwidth in a much smaller board footprint
- Highly parallel connections to memory stacks inside the package for fast access
- Significant reduction in power consumption
“This design greatly expands the possibilities for deep learning, and we are delighted to enter final qualification,” said Ajay Lalwani, vice president, global manufacturing operations at eSilicon. “TSMC’s 2.5D CoWoS packaging technology has been a key differentiator for this advanced design.”
About eSilicon
eSilicon is an independent provider of complex FinFET-class ASICs, custom IP and advanced 2.5D packaging solutions. Our ASIC+IP synergies include complete 2.5D/HBM2 and TCAM platforms for FinFET technology at 14/16/7nm as well as SerDes, specialized memory compilers and I/O libraries. Supported by patented knowledge base and optimization technology, eSilicon delivers a transparent, collaborative, flexible customer experience to serve the high-bandwidth networking, high-performance computing, artificial intelligence (AI) and 5G infrastructure markets. www.esilicon.com
Collaborate. Differentiate. Win.™
eSilicon is a registered trademark, and the eSilicon logo and “Collaborate. Differentiate. Win.” are trademarks, of eSilicon Corporation. Other trademarks are the property of their respective owners.
| Contacts: | |
| Sally Slemons | Susan Cain |
| eSilicon Corporation | Cain Communications |
| 408-635-6409 | 408-393-4794 |
| [email protected] | [email protected] |


OpenAI GPT-5.6 Set for Wider Release After U.S. Commerce Approval, Report Says
Nvidia Invests $500M in Firmus Technologies Ahead of Planned ASX IPO
SK Hynix’s $28 Billion U.S. Share Sale Draws Massive Demand Amid AI Chip Boom
BHP Faces Port Hedland Strike Threat as Iron Ore Export Risks Grow
Fast Retailing Raises Full-Year Forecast After Uniqlo Owner Beats Q3 Profit Estimates
SK Hynix’s $28B U.S. IPO Draws Strong Demand as AI Chip Boom Fuels Investor Interest
Bain Capital Exits Kioxia After AI-Fueled Valuation Surge
Telenor to Buy Controlling Stake in Bahnhof in $630 Million Broadband Deal
SK Hynix Soars 13% in Nasdaq Debut After Record $26.5 Billion IPO
Levi Strauss Raises 2026 Outlook After Q2 Earnings Beat, Shares Drop Despite Strong Results
OpenAI Executive Fidji Simo to Step Down Amid Health Challenges Ahead of IPO
Mizuho’s Top U.S. Industrials Stocks: Why Corteva and Stanley Black & Decker Stand Out
Kitron Q2 Revenue Beats Estimates as Defense Demand Lifts Growth
Apple Sues OpenAI, Former Employees Over Alleged Trade Secret Theft
Oppenheimer Sees CNH Industrial as Top 2026 Agriculture Stock Pick on Dealer Consolidation Strategy
Japan Regional Bank Stocks Drop After Zentoshin Bankruptcy Sparks Credit Risk Concerns
Goldman AM Sees Strong Buyout Opportunities in Japan, South Korea and Australia 



