Intel's 3D XPoint Plans Clearer Than Micron's

发布时间:2018-07-20 00:00
作者:Ameya360
来源:EE Times
阅读量:1807

Or who you don't talk to. Micron, for its part, isn't offering any more guidance right now beyond what was stated in a joint news release issued earlier this week. "The companies have agreed to complete joint development for the second generation of 3D XPoint technology, which is expected to occur in the first half of 2019," the statement reads. "Technology development beyond the second generation of 3D XPoint technology will be pursued independently by the two companies in order to optimize the technology for their respective product and business needs."

Intel is still bullish on the technology. In a telephone interview with EE Times, Bill Leszinske, vice president of Intel's non-volatile memory solutions group, said it makes sense for Intel to continue on its present path.

"We believe in the technology," Leszinske said. He added that what Intel now calls Optane technology presents opportunities on a DIMM in a server as well as high-performance storage either in a server or a client device.

Not surprisingly, Leszinske wouldn't speak on behalf of Micron or speculate on its intentions but did say that Intel believes in the 3D XPoint technology direction the partners had jointly chosen. “Just like in 3D NAND, we believe in the floating gate direction that we've chosen," he said. "In the end, Micron said they have a different direction that they would like to head."

The biggest impact of the Intel/Micron partnership wind-down is that each company will have to completely fund its own technology development, said Leszinske. He acknowledged that this is an expensive commitment, one that includes putting Optane on DIMMs — announced by Intel in May. Dubbed Optane DC Persistent Memory, it's being sampled by customers now, he said, and will be in production the second half of this year, followed by broader availability in 2019.

So far, 3D Xpoint has only found its way into SSDs. As storage media, it may become what NAND flash was in its early days — an augmentation to spinning disk-based storage for data that needs to be accessed frequently and quickly, with flash taking the place of hard drives. "For us, having Optane and low-cost 3D NAND is our direction," said Leszinske. "We have the ability to go invest in that, and we will."

Optane is all about high performance/low latency, he said, while NAND is all about high-density affordable costs so Intel can replace hard drives and reduce customers' operating costs in a data center or make it a better-performing client device. "We will shoot for lower cost even if the performance isn't quite as good because we think bulk replacing of hard drives and building out more tiers of salvaging storage is a better business result for our customers," Leszinske said. 

Intel's Optane has found its way into real-world use cases. The University of Pisa in Italy, for example, is able to process MRIs in two minutes rather than 40, something that claustrophobic patients appreciate, said Leszinske. "That's been an amazing experience that we never would have envisioned. We know there's more of those things out there," he said. 

Intel's enthusiasm for 3D XPoint sounds every much like the buzz around the initial announcement that it and Micron were developing an entirely new class of non-volatile memory that claimed dramatically lower latency and exponentially greater endurance than NAND memory. At the time, there was a lot of speculation about 3D XPoint and the expectation that it would turn up in SSDs and DIMMs.

It's been nearly three years since 3D XPoint was unveiled. As Jim Handy, principal analyst at Objective Analysis, notes, it didn't take long to figure out that the SSDs weren't going to be able to take advantage of 3D XPoint's speed.

"And they seem to be in a struggle to get the DIMM out for whatever reason," Handy said, adding that Micron recently said that its Lehi, Utah, facility was idle. "And the only thing that the Lehi plant has been making for the past six months is 3D XPoint memory," Handy said. Objective Analyst notes that Intel's NAND flash business has been underperforming while other plays are doing quite well.

Handy said that this would seem to suggest that there's over-inventory of 3D XPoint SSDs and that they aren't selling as well as they could. "The SSDs are not really compelling," Handy said. "They're significantly more expensive and nominally better-performing than a NAND flash SSD."

Handy said that it's important to note that Intel is the only one selling anything with 3D XPoint in it. It's also important to remember that Micron and Intel's businesses are different.

"Micron's in an interesting place because they know exactly what it cost to make 3D XPoint memory and they have chosen not to introduce SSDs," said Handy. "Which, to me, says that they don't want lose money on it like Intel is." Micron would have to answer to investors if they lose money on the technology, but for Intel, losing money is fine as long as 3D XPoint helps with its Xeon sales, Handy said.

That being said, Intel's storage unit has been underperforming while everyone else in the NAND flash business "has been reaping in gobs of money," said Handy. And although Intel may optimistic about the opportunities for 3D XPoint, Intel being the only supplier of the technology might put a damper on plans by OEMs such as Dell or HP to build a product line around Optane. "How warm and fuzzy would you feel about your source of supply?"

(备注:文章来源于网络,信息仅供参考,不代表本网站观点,如有侵权请联系删除!)

在线留言询价

相关阅读
Intel’s Advanced Packaging, Led by EMIB-T, Set to Hit Its Stride in 2029; DRAM Manufacturing Ruled Out
  Intel is gaining momentum in advanced packaging, with tech giants such as Google reportedly showing interest in its EMIB (Embedded Multi-die Interconnect Bridge) technology amid ongoing CoWoS supply constraints. Citing CFO David Zinsner’s remarks at Deutsche Bank’s 2026 Technology Conference on August 26, Investing.com reports that Intel expects advanced packaging revenue, led primarily by EMIB-T, to ramp in 2H27, become a steady contributor in 2028 and hit its stride in 2029.  Memory was another key topic at the conference, as AI reshapes the strategic importance of memory. Tom’s Hardware previously reported that Intel CEO Lip-Bu Tan had hinted at a potential return to the memory market. Zinsner, however, clarified that Intel has no plans to re-enter DRAM manufacturing, according to Investing.com.  Advanced Packaging Eyes 40% Gross Margin  Notably, Zinsner highlighted the scale and profitability potential of Intel’s expanding advanced packaging business. According to Investing.com, the opportunity could be worth several billion dollars per customer annually, with the business targeting around 40% gross and 30% operating margins. Its relatively low capital intensity versus front-end manufacturing could also support strong ROIC, he added.  In Zinsner’s words, Intel is “already eating our own dog food” with EMIB, indicating that the company is already deploying the technology in its own products. Building on EMIB, EMIB-T eliminates the need for an interposer, further enhancing the technology’s appeal and potential for broader adoption, he said.  According to Wccftech, Intel’s Clearwater Forest, or Xeon 6+, CPU features twelve EMIB tiles using 2.5D packaging. The design integrates three active base tiles with two I/O tiles and twelve compute tiles. The I/O tiles are built on Intel 7, while the active base tiles use Intel 3 and the compute chiplets are manufactured on Intel 18A, the report explains.  SK hynix Veteran’s Appointment Signals Intel’s Packaging Push  Lip-Bu Tan’s appointment of former SK hynix CEO Seok-Hee Lee as Intel Foundry’s executive vice president could fuel speculation that Intel is considering a return to memory manufacturing. However, according to an Investing.com transcript, Intel CFO David Zinsner instead pointed to Lee’s appointment as a strong vote of confidence in the growth potential of EMIB-T.  Bringing in an executive of Lee’s caliber to lead the business underscores Intel’s confidence in the opportunity and its long-term prospects, he added.  Intel Stays Out of DRAM Manufacturing  Notably, while Intel is developing memory-related technologies such as XHBM and Z-angle Memory (ZAM), Zinsner pushed back on speculation that the chipmaker plans to return to memory manufacturing.  According to Investing.com, he said Intel has no plans to resume DRAM production. Instead, Intel is working with three major memory players on AI-focused products and architectures designed to ease memory bottlenecks. Management emphasized that the company’s focus will remain on system-level memory solutions rather than manufacturing memory itself, according to the Investing.com transcript.
2026-08-31 16:02 阅读量:293
Intel Patent Reveals XBM Matching HBM4 Footprint Without Interposers; Commercialization Seen After 2030
  As AI chip demand surges, existing high-bandwidth memory (HBM) is facing supply and cost challenges, prompting the search for alternative memory technologies. According to Tom’s Hardware, an Intel patent application published on July 2, 2026 reveals the company’s proposed cross-batch memory (XBM) architecture, designed to ease the packaging and cost bottlenecks of today’s interposer-based HBM. The report states that the design aims to match HBM4’s footprint while replacing conventional DRAM and its ultra-wide interface with back-end-of-line (BEOL) transistors and serial Universal Chiplet Interconnect Express (UCIe) links.  As noted by Wccftech, XBM is expected to target commercialization after 2030, in line with ZAM, the memory architecture Intel is co-developing with SoftBank subsidiary SAIMEMORY.  XBM Architecture and Advantages  The proposed architecture centers on DRAM blocks connected to a UCIe I/O block operating at 32 GT/s, with the I/O routed through the base die. According to Wccftech, each XBM stack provides a die capacity ranging from 0.5GB to 5GB. Each sub-channel consists of 12 data blocks, with up to 96 data blocks in an 8-high XBM stack and 192 in a 16-high stack. These channels operate at 2GHz. The report also notes that XBM can be implemented in multiple package configurations, including Memory-on-Package (MoP), enabling higher bandwidth and capacity in smaller form-factor designs.  Notably, the memory die uses 1T1C (one transistor, one capacitor) back-end DRAM, with transistors fabricated in the back-end-of-line (BEOL) metal layers rather than front-end silicon, Wccftech says. According to the report, this significantly improves area efficiency, allowing more space for TSVs (Through-Silicon Via) and enabling higher memory density and bandwidth.  Competitive and Ecosystem Challenges  The global HBM market is currently dominated by South Korean suppliers. As Global Economic News points out, conventional HBM incurs high manufacturing costs due to the micro-bump processes used to vertically stack DRAM dies, while silicon interposers add significant routing complexity and cost. Intel’s proposed XBM architecture is designed to address these limitations.  However, Global Economic News also notes that Intel’s proposal is unlikely to immediately shift the industry’s competitive landscape. SK hynix and Samsung Electronics have spent several years developing cost-saving technologies, including standard chiplets, UCIe, and fan-out packaging, to reduce interposer costs. The report further notes that platform compatibility and software ecosystems remain major barriers to adoption. The global AI accelerator ecosystem, led by NVIDIA, is currently optimized for the existing HBM architecture and wide-bandwidth parallel interfaces, making a transition to alternative memory architectures more challenging.
2026-07-09 11:24 阅读量:587
Intel Reportedly Weighs Dual-Side Power Delivery for 14A2 to Boost Chip Density in Race With TSMC, Samsung
  Intel is reportedly refining its 1.4nm roadmap with a new power delivery architecture. According to ETNews, sources say the company originally planned to rely exclusively on its PowerDirect back-side power delivery network (BSPDN) for the baseline 14A node. However, Intel is now reportedly considering a dual-side architecture for the follow-on 14A2 process that combines both front-side and back-side power delivery.  The report says the architectural change is driven by lithography limitations as Intel pushes the pitch of its lowest metal layer (M0) to around 21nm, where stochastic defects become increasingly difficult to manage.  Intel has previously said it aims to increase chip density by 1.3× over 18A to compete with TSMC’s N2/A14 and Samsung’s SF2Z. While 14A targets an M0 pitch of approximately 28nm, industry analysis cited by the report suggests the follow-on 14A2 could reduce it to around 21nm. The report says the resulting density gains would help justify the use of High-NA EUV lithography systems, despite requiring double patterning, which cost hundreds of billions of Korean won.  The challenge, however, is that shrinking circuit lines below 21nm causes interconnect resistance to rise exponentially. The report says the nano Through-Silicon Via (nTSV) infrastructure built for back-side power delivery alone may no longer meet the current density requirements of the transistors, leading to severe IR drop, or voltage loss. Intel is therefore expected to keep the back-side power delivery network as the primary power path while repurposing portions of the front-side metal layers to carry supplemental power and clock signals.  The report says the hybrid architecture is a design compromise aimed at preserving sufficient power margin as Intel pushes toward more aggressive scaling while confronting lithography limitations. Although it adds routing complexity, the approach is viewed as necessary to achieve the targeted 21nm M0 pitch.  Intel’s reported move also comes as TSMC advances its Super Power Rail (SPR) backside power delivery roadmap, with A16 slated for volume production in 2027 and A12 targeted for 2029, both incorporating the technology for AI and HPC applications, according to Tom’s Hardware.  Roadmap Comparison  The ETNews report also highlights Intel’s tight development timeline. According to its roadmap, the 14A process is scheduled to enter risk production in 2028, followed by volume production in 2029. Intel is also expected to distribute version 0.9 of its 14A Process Design Kit (PDK) to external customers in October, the report adds.  By the time Intel reaches risk production, however, TSMC is expected to already be shipping commercial products built on its A14 process. The report notes that TSMC had already achieved stable yields on its N2 process during 2025–2026 and entered the market in line with Apple’s product launch schedule. As for Samsung, The Elec reports that the company has recently reaffirmed plans to begin mass production of its 1.4nm process in 2029, followed by the enhanced SF1.4+ node in 2030.
2026-07-07 11:21 阅读量:692
Intel Taps SK hynix Ex-CEO as Foundry EVP to Lead Advanced Packaging Push as EMIB, HBI Scale Up
  Intel CEO Lip-Bu Tan is accelerating the company’s foundry ambitions, continuing to recruit seasoned talent from across the chip industry. The latest move is the appointment of Seok-Hee Lee as Executive Vice President of Intel Foundry, according to an Intel announcement on June 18.  Notably, Lee joins Intel after serving as president and CEO of SK On and previously leading SK hynix. A semiconductor industry veteran, he also held engineering leadership roles at Intel and in academia, bringing deep expertise in advanced process technologies and large-scale manufacturing, according to the press release.  Reporting directly to Tan, Lee will oversee advanced packaging, system integration, back-end technology development, and manufacturing, as noted by the company. Intel said the appointment is aimed at strengthening its system-level innovation capabilities and supporting the next phase of its foundry expansion.  Tan also underscored Lee’s pivotal role in advancing Intel’s next-generation packaging strategy.“As we prepare to scale advanced packaging technologies, including EMIB-T and HBI, for high-volume production, Seok-Hee is the ideal leader to drive the growth of this strategically important business,” Tan said.  The appointment, as Reuters points out, comes after U.S. President Donald Trump announced earlier that Apple had agreed to collaborate with Intel on designing and manufacturing chips in the United States, providing a potential boost to Intel’s foundry business.  A previous Reuters report, citing The Information, also reported that Google has tapped Intel to manufacture more than 3 million TPUs in 2028, while NVIDIA is evaluating Intel’s 18A process and advanced packaging technologies for a next-generation multi-die GPU design.  Meanwhile, Intel Foundry EVP Naga Chandrasekaran will continue reporting to CEO Lip-Bu Tan, overseeing front-end technology development and manufacturing as Intel accelerates the ramp of Intel 18A, Intel 14A and future process nodes. He will also retain responsibility for design enablement and customer-facing business operations, supporting Intel Foundry’s long-term growth.  Intel’s Talent Offensive Continues  Lee’s appointment marks the latest in a series of high-profile hires since Lip-Bu Tan took the helm at Intel. Reuters notes that the company bolstered its foundry ambitions in April by recruiting Samsung foundry veteran Shawn Han to support its contract manufacturing business.  The hiring drive began earlier. In late 2025, Intel brought in former TSMC senior vice president Wei-Ren Lo as executive vice president overseeing manufacturing and packaging amid controversy surrounding alleged sub-2nm trade secrets. Intel said Lo’s return was part of its broader transformation effort, with the veteran executive bringing back nearly two decades of experience gained at Intel before joining TSMC.
2026-06-22 10:51 阅读量:632
  • 一周热料
  • 紧缺物料秒杀
型号 品牌 询价
TL431ACLPR Texas Instruments
RB751G-40T2R ROHM Semiconductor
CDZVT2R20B ROHM Semiconductor
MC33074DR2G onsemi
BD71847AMWV-E2 ROHM Semiconductor
型号 品牌 抢购
IPZ40N04S5L4R8ATMA1 Infineon Technologies
TPS63050YFFR Texas Instruments
BU33JA2MNVX-CTL ROHM Semiconductor
BP3621 ROHM Semiconductor
ESR03EZPJ151 ROHM Semiconductor
STM32F429IGT6 STMicroelectronics
热门标签
ROHM
Aavid
Averlogic
开发板
SUSUMU
NXP
PCB
传感器
半导体
相关百科
关于我们
AMEYA360微信服务号 AMEYA360微信服务号
AMEYA360商城(www.ameya360.com)上线于2011年,现 有超过3500家优质供应商,收录600万种产品型号数据,100 多万种元器件库存可供选购,产品覆盖MCU+存储器+电源芯 片+IGBT+MOS管+运放+射频蓝牙+传感器+电阻电容电感+ 连接器等多个领域,平台主营业务涵盖电子元器件现货销售、 BOM配单及提供产品配套资料等,为广大客户提供一站式购 销服务。

请输入下方图片中的验证码:

验证码