While making strides in the HBM capacity expansion at its Cheongju M15X fab, SK hynix may have locked in another major client in addition to NVIDIA. According to South Korean media outlet The Elec, the memory giant has reportedly landed a substantial order to provide HBM to Broadcom.

Sources cited by The Elec indicate that SK hynix is set to deliver HBM to Broadcom in the latter half of 2025, and the U.S. chipmaker will likely use the memory chips in AI computing solutions for a Big Tech client.

SK hynix initially planned to increase its 1b DRAM production capacity to 140 – 150K 300mm wafers per month by 2025, as the report noted. However, the recent deal with Broadcom is expected to push this target higher, reaching 160 – 170K 300mm wafers per month.

Notably, The Elec notes that SK hynix may tend to postpone the installation of its 1c DRAM equipment, the successor to 1b DRAM, to address the surge in demand driven by Broadcom.

Major CSPs have become increasingly ambitious on developing their own AI infrastructures, thus driving the demand for AI-specific ASICs and HBM. For instance, Google has reportedly integrated HBM3E into Trillium, its self-developed 6th-generation TPU, as AWS also utilizes HBM in its self-developed Trainium chipset for AI learning purposes.

Earlier this month, Broadcom announced it is collaborating with three major cloud service providers – likely Google, Meta, and ByteDance – on the development of AI chips, as noted by The Elec.

Read more

• [News] NVIDIA CEO Jensen Huang Reportedly Asked SK hynix to Expedite HBM4 Supply by 6 Months
• [News] SK Hynix Reportedly Expands HBM Production at Cheongju M15X Fab, Aiming for Late 2025 Completion

(Photo credit: SK hynix)

Amid the booming demand for advanced packaging used in AI chips, TSMC Chairman C.C. Wei confirmed in the July earnings call that the company has been working on a mini production line for FOPLP (Fan-Out Panel Level Packaging), with achievements expected within three years.

Now here’s the latest update: The Taiwanese foundry giant has reportedly set an initial specification for the FOPLP substrate size, as it prefers the 300x300mm size instead of the rumored 515x510mm, as per Taiwanese media outlet MoneyDJ.

Equipment suppliers are said to be working closely on the development of the technology, with the mini line expected to be set up by 2026 at the earliest and ramping up gradually in 2027, the MoneyDJ report indicates.

According to industry sources cited by MoneyDJ, TSMC initially favored a rectangular substrate size of 515mmx510mm. However, it chose to adopt the 300mx300mm eventually because of the “Cost of Ownership” (COO) factor and the maximum reticle size that can be supported.

As larger substrate sizes not only exacerbate issues with warping but also lead to higher risks of damage during transportation and packaging process transitions, TSMC is said to start with the 300x300mm size, the MoneyDJ report notes, adding that it may look to increase the PLP packaging substrate size as technologies progress in the future.

TSMC introduced the FOWLP (Fan-Out Wafer Level Packaging) technology named InFO (Integrated Fan-Out) in 2016, first used in the iPhone 7’s A10 processor. Subsequently, the demand for FOPLP solutions is on the rise, as customers are attracted to more cost-effective solutions. The rising global demand for AI chips, which continues to pressure advanced packaging capacity, is also fueling the need for FOPLP.

In addition to TSMC, Intel and Samsung have also been advancing in this area. Samsung currently offers advanced packaging services such as I-Cube 2.5D packaging, X-Cube 3D IC packaging, and 2D FOPKG packaging. For applications requiring low-power memory integration, such as mobile phones or wearable devices, Samsung already provides platforms like fan-out panel-level packaging and fan-out wafer-level packaging.

Intel, on the other hand, is planning to launch the industry’s first glass substrate solution for next-generation advanced packaging, with mass production scheduled between 2026 and 2030.

Read more

• [News] Packaging and Equipment Firms Accelerate FOPLP Deployment: Spotlight on 7 Taiwanese Companies
• [News] TSMC and Samsung: Close Combat on FOPLP, with Different Panel Materials for Packaging

(Photo credit: TSMC)

On December 18, Japan’s leading NAND Flash manufacturer, Kioxia, debuted on the Tokyo Stock Exchange’s Prime Market with its Initial Public Offering (IPO). According to a report from MoneyDJ, citing Nikkei, Kioxia’s president, Nobuo Hayasaka, announced on December 19 that the company plans to urge its two major shareholders, Bain Capital and Toshiba, to sell additional shares to meet the requirements for continued listing.

Plans to Meet Requirements for Continued Listing

According to the Nikkei report on December 19, companies listed on the Tokyo Stock Exchange’s Prime Market must ensure that at least 35% of their shares are publicly traded within five years. Currently, Kioxia’s publicly traded share ratio stands at 28%. To fulfill the requirements for continued listing, the company intends to urge its two major shareholders, Bain Capital and Toshiba, to sell additional shares within the next three years, aiming to reach the 35% threshold by 2030.

Notably, following its IPO, HBM giant SK hynix has become Kioxia’s third-largest shareholder, holding a 14% stake, according to another Nikkei report.

Focus on NAND Flash for AI Applications

In addition to outlining plans to meet the requirements for continued listing, Hayasaka also emphasized Kioxia’s objective to enhance profitability through NAND Flash products  for AI applications, according to the report. The company aims to double its current market share in this segment, targeting an increase from 9% to 17% within the next 2–3 years.

The report also highlights Kioxia’s plans to to ramp up production of its cutting-edge NAND Flash products at its Kitakami plant, scheduled to begin in September 2025. These next-generation NAND Flash products are projected to deliver an 80% increase in data transfer speeds.

According to TrendForce, in NAND Flash market, Kioxia ranked third in revenue in the third quarter of 2024, with a 15.1% market share, after Samsung (35.2%) and SK Group (20.6%)

Beyond NAND Flash, the company is also working on next-generation memory technologies. According to the report, citing Hayasaka, it takes at least three years from research and development to the production of new products, and Kioxia is currently developing a new type of DRAM using oxide materials, which aims to enhance energy efficiency.

Read more

• [News] Kioxia Finalizes IPO Price, SK Hynix Poised to Become Third-Largest Shareholder
• [News] Kioxia Reportedly Planning Production Cuts Amid NAND Flash Market Challenges

(Photo credit: Kioxia)

According to a report from Reuters, citing The Information, the U.S. Department of Commerce has recently requested NVIDIA to investigate how its AI chips ended up in China over the past year. This request comes as part of ongoing efforts to enforce export control regulations and address potential unauthorized resales.

NVIDIA Ordered Spot Checks for Distributors

The report notes that NVIDIA has asked major distributors, including Super Micro Computer and Dell Technologies, to perform spot checks on their clients in Southeast Asia, since servers from Super Micro and Dell are integrated with NVIDIA AI chips.

Smuggling Activities and Forged Server Serial Numbers

Despite these measures, five individuals involved in smuggling NVIDIA chips claimed they have successfully evaded detection during recent inspections conducted by Super Micro, according to The Information.

The report also highlights, based on a source close to Super Micro, that some customers duplicated the serial numbers of servers equipped with NVIDIA chips purchased from Super Micro and applied them to other servers within their access. In some cases, smugglers reportedly altered the serial numbers within the servers’ operating systems to avoid detection.

Response from Super Micro and Dell

In response to the investigation, Dell Technologies reaffirmed its commitment to compliance, stating that it requires distributors and resellers to follow all applicable regulations. Dell also noted that it might terminate relationships with non-compliant partners. Similarly, Super Micro stated that it investigates and takes action against any unauthorized exports or re-exports of its products by third parties, adhering to U.S. export control requirements under the Export Administration Regulation.

NVIDIA AI Chips Sales in China Despite Restrictions

The report points out that the Biden administration has intensified its crackdown on AI chip exports to China. Despite these restrictions, several Chinese universities and research institutes have managed to acquire NVIDIA chips through resellers, according to Reuters.

A Wccftech report from April 2024 also notes that a Shenzhen merchant is promoting the sale of the NVIDIA H100 SXM5 8 GPU Xeon AI Supermicro server, equipped with Nvidia H100 chips on the Chinese IT website “ZOL.”

NVIDIA’s call for a probe into its distributors comes at a time when its relationship with Super Micro appears to be under strain. According to a report from Liberty Times from November, citing Chinese media outlet Cailian Press, NVIDIA has reportedly been shifting orders that were previously directed to Supermicro to other suppliers to avoid risks caused by Super Micro’s financial troubles since the company was under misconduct allegations at that time.

Read more

• [News] China Opens Antitrust Probe Against NVIDIA, Fines Could Hit $5 Billion
• [News] NVIDIA Rumored to Shift Supermicro Orders, Potentially Benefiting Taiwanese Suppliers

(Photo credit: NVIDIA)

After finalizing the USD 7.86 billion funding award under U.S. CHIPS Act, Intel may be close to another business turnaround. According to Bloomberg, Team Blue has advanced several buyout firms to the next stage of bidding for Altera, its FPGA unit.

A deadline for potential buyers to submit formal offers has also been set, as they need to be provided by the end of January, according to sources cited by the report.

Bloomberg notes that private equity firms such as Francisco Partners and Silver Lake Management are among the contenders, while Apollo Global Management and Bain Capital are also in the running for Altera.

Lattice: the Only Semiconductor Contender Surfaced by Now

Notably, Lattice Semiconductor, one of Altera’s FPGA (Field Programmable Gate Array) peers, is also rumored to be competing in the second round, as per Bloomberg. For now, it remains the only tech company surfaced to show interests in the deal.

Gaining control of Altera could be a challenging endeavor for Lattice. As per Bloomberg, with a market value of USD 8 billion, Lattice may need additional financial support.

Representatives from the aforementioned parties declined to comment, Bloomberg notes.

Following its request for initial bids before Thanksgiving in November, Intel has reportedly received various deal-structure proposals, ranging from acquiring a 20% to 30% stake in Altera to seeking full ownership, according to sources cited by Bloomberg.

Moreover, Bloomberg indicates that some bidders have presented multiple options in their proposals, with valuations for the unit ranging from USD 9 billion to over USD 12 billion. The scale of the offers, however, is significant lower than the USD 17 billion Intel paid to acquire Altera in 2015.

The “IMS” Model?

Intel’s interim co-CEO David Zinsner suggested that the company might be inclined to bring in another partner, similar to what they did with the IMS business, the Bloomberg report adds.

According to Intel’s press release, IMS is an industry leader in multi-beam mask writing tools required to develop advanced extreme ultraviolet lithography (EUV).

In June 2023, Intel sold a 20% stake in its IMS Nanofabrication business to Bain Capital, with the deal valuing the unit at around USD 4.3 billion. Three months later, an additional 10% stake in IMS was sold to Taiwanese foundry giant TSMC.

Read more

• [News] U.S. FPGA Peer Lattice Reportedly Explores Acquisition of Intel’s Altera Unit
• [News] Intel Mulls Sale of Altera Stake, while Silver Lake and Bain Reported as Potential Buyers

(Photo credit: Intel)

According to a report from India Today, a post by “Fixed Focus Digital” on the Chinese platform Weibo reveals that Apple might abandon its iPhone SE branding and rebrand it as “iPhone 16e.” Rumor has it that the smart phone is likely to be more aligned with the iPhone 16, such as in its camera and display specifications, representing a significant upgrade from its predecessors, as noted in the report

The rumored A18 chipset is expected to give the iPhone 16e a significant performance boost, bringing it closer to the iPhone 16 series in processing capabilities, as noted by the report. Notably, the increase to 8GB of RAM (up from 4GB) is particularly significant, enhancing multitasking capabilities and supporting Apple Intelligence, according to the report.

Additionally, the Weibo tipster reveals that the iPhone 16e will share the same screen size as the standard iPhone 16 and feature a full-screen design. The report mentions that the upcoming iPhone 16e is expected to feature a 6.06-inch OLED display, a notable upgrade from the 4.7-inch LCD screen of its predecessor.

The report also notes that replacing the Touch ID home button with Face ID is another significant change, aligning the phone’s design more closely with Apple’s latest models.

Regarding pricing, the report suggests it could range from USD 499 to USD 549, marking an increase from the iPhone SE 3’s starting price of USD 429.

Meanwhile, rumors about Apple discontinuing the SE series align with reports that the company plans to stop selling the iPhone SE and iPhone 14 series in Europe by the end of the year. According to a report from 9to5Mac, these models feature a Lightning port, which do not comply with the European Union’s policy taking effect in January 2025.

Read more

• [News] Apple Reportedly Set to Launch AI-Powered iPhone SE in Early 2025, Boosting Key Suppliers
• [News] Apple Reportedly Planning 2028 Launch for Foldable iPad as Large as Two iPad Pros

(Photo credit: Apple)

TSMC and Samsung are not just archrivals in the foundry market. They also confront each other in an emerging sector, the Fan-Out Panel-Level Packaging (FOPLP) technology. According to South Korean media outlet ETnews, the two semiconductor giants, while both eyeing FOPLP as a secret weapon to high-performance computing packaging for AI, approach the tech with different core materials.

According to a previous report from Commercial Times, TSMC Chairman C.C. Wei confirmed that the company has established an FOPLP R&D team and production line, with achievements expected to be seen within three years.

On the other hand, Business Korea noted that Samsung is making strides in the PLP field, as the tech giant acquired the business from Samsung Electro-Mechanics in 2019.

The two companies reportedly adopt different panel materials for FOPLP. As per ETnews, Samsung is expected to choose plastic (organic) as the core material, which is commonly used in PCBs. In contrast, TSMC would focus on using glass as the panel material.

Samsung: A Veteran in Plastic Panels

While the panel plays a crucial role in supporting the semiconductor die and facilitating the packaging process, the choice between plastic and glass will result in differences in performance and processing, which could lead to variations in the outcomes of their packaging businesses, as per ETnews.

To elaborate a bit, the ETnews report notes that FOPLP uses square panels instead of round wafers, allowing it to eliminate chip waste at the corners. For example, with a 600×600mm panel, the packaging method is possible to produce 5-6 times more chips than with a 300mm (12-inch) wafer, the report suggests.

Given that Samsung has utilized plastic panels on packaging mobile APs and PMICs, it is likely to stick with the material when applied FOPLP to AI chip packaging, the report adds, noting that the company has reportedly begun related FOPLP research and development under the leadership of the AVP development team last year.

TSMC: Teams up with Innolux on Glass Panels

In contrast, TSMC is considering glass panels on FOPLP. Though the Taiwanese foundry giant has been using CoWoS (Chip-on-Wafer-on-Substrate) 2.5D packaging technology to package AI semiconductors for clients like NVIDIA, it is preparing for FOPLP due to production capacity limitations and cost concerns, the ETnews report notes.

TSMC’s acquisition of Taiwanese panel manufacturer Innolux’s plant in August, thus, is regarded as a critical move to advance in FOPLP production. The two companies are likely to collaborate with on glass panel technology, as per the report.

The report notes that using glass for FOPLP panels is considered advantageous in terms of reducing issues like foreign particles and panel warping. However, handling glass is also regarded as challenging, making the process more difficult.

According to the analysis by TrendForce, the main advantages of FOPLP are lower unit costs and larger packaging sizes, but the technology and equipment systems still need development, and the commercialization process is highly uncertain.

The estimated mass production timeline for FOPLP packaging technology in consumer IC and AI GPU applications is the second half of 2024 to 2026 and 2027–28, respectively, as per TrendForce. Is it TSMC or Samsung that prevails in the FOPLP battle? Only time will tell.

Read more

• [News] TSMC Pushes for FOPLP Mass Production by 2027, Reportedly Eyeing on Innolux’s Plant
• [News] TSMC Reportedly Forms a Team on FOPLP Development, with Mini Line on the Road

(Photo credit: Samsung)

JBD, a pioneering MicroLED display manufacturer, has set a new standard with its Phoenix series microdisplay, achieving an industry-record white-balanced brightness of 2 million nits.

Leveraging revolutionary coaxial stacking architecture and sophisticated manufacturing techniques, JBD has achieved a monumental leap in brightness, significantly amplifying the Phoenix series’ potential for immersive augmented reality experiences.

Since its introduction in 2023, the Phoenix series has demonstrated industry-leading capabilities:

• Unrivaled Pixel Density: With a RGB-Combined pixel pitch of 5μm, it delivers exceptional resolution on tiny displays.
• Proprietary Coaxial stacking: Features an ultra-thin stack thickness of less than 5μm for enhanced efficiency.
• Native Semiconductor Emitters: Built on inorganic III-V and III-Nitride compound semiconductor materials.
• Exceptional brightness: Doubled brightness from 1 million to 2 million nits in a year.
• Vivid color fidelity: < 25nm FWHM spectrum width ensuring exceptional color purity and accuracy.

The Phoenix series, with its monolithic design, streamlines projector architecture, enabling an expansive field of view over 50 degrees, ideal for immersive AR applications. These innovations support immersive, all-day AR experiences, offering unmatched flexibility for diverse AR applications.

Building on these advancements, JBD is set to commence mass production of the Phoenix series in Q3 2025, focusing initially on deliveries to selected customers with customization request partners. This milestone marks an important step forward in the commercialization of its cutting-edge Native Monolithic RGB MicroLED technology.

JBD offers a diverse portfolio of MicroLED polychrome solutions, with significant progress made in 2024 through its X-cube solutions. The Hummingbird I polychrome projector now achieves a luminous flux at least 6 lumens, while its latest optical module delivers 6,000 nits of brightness with 30 degrees FOV diffractive waveguides—paving the way for consumer-grade color AR glasses.

(Photo credit: JBD)

TrendForce 2024 Near-Eye Display Market Trend and Technology Analysis
Release Date:2024 / 07 / 31
Languages:Traditional Chinese / English
Format:PDF
Page:164

TCL Electronics, a leading consumer electronics brand and one of the world’s top two TV brands, today announced its attendance at Consumer Electronics Show (CES) 2025 which takes place in January. TCL has participated in CES for 31 years, during which the brand has consistently showcased its latest innovative display technologies and a full range of smart home products to global visitors.

As a highlight of this year’s showcase, TCL will reveal its innovative QD-Mini LED technologies and products that deliver an unparalleled visual experience. The display innovations TCL is bringing to CES 2025 are sure to excite both professionals and home entertainment enthusiasts alike. They cover almost every form factor, from professional monitors to in-car displays, mobile phones, tablets, smart watches, smart projectors, and RayNeo AR glasses. TCL will also display its award-winning NXTPAPER technology and officially unveil its revolutionary NXTPAPER 4.0 at CES 2025.

Underscoring its innovations in the AI field, TCL is bringing its latest AI advancements in new product categories as well as partnerships with world-renowned technology brands. Additionally, TCL will unveil the integration of its smart home ecosystem with intelligent solutions designed to enhance energy efficiency and enrich the user experience, which includes air conditioners, built-in refrigerators, washing machines, smart locks, and mobile routers.

Leading up to Super Bowl LIX, TCL’s partnerships with renowned global partners, including its role as the Official TV Partner of the National Football League (NFL) in the United States, will also be a major highlight of the exhibition. Through these partnerships, TCL is developing cutting-edge technologies to enhance viewer experiences and engage fans across a range of interactive on-site activities and online campaigns that aim to build global connections and inspire greatness in everyday life.

(Photo credit: TCL)

TrendForce 2025 Mini LED Backlight Display Trend and OLED Technology Competition Analysis
Release Date:
PDF (194 Pages)- 28 August 2024
Languages: Traditional Chinese / English

As TSMC keeps advancing in the most cutting-edge nodes, its foundry peers in Taiwan may have to deal with multiple challenges. In addition to the price war launched by their Chinese rivals, Taiwanese foundries focusing on legacy nodes, including UMC and TSMC’s affiliate VIS, will face significantly increased depreciation pressure starting from 2025, as per a report from Commercial Times.

According to the report, amid their ongoing expansion efforts, major mature process foundries in Taiwan will face increased depreciation pressure afterwards. Sources cited by the report suggest that UMC’s depreciation cost is expected to increase by more than 20% year-on-year in 2024, while its depreciation in 2025 may reach a new high.

In the third quarter, UMC’s depreciation expenses reached NTD 12.059 billion (roughly USD 369 million), a near six-year high. Its accumulated depreciation for the first three quarters of 2024, on the other hand, totaled NTD 32.802 billion (roughly USD 1 billion), a year-on-year increase of 18%, as noted by the report.

The report indicates that UMC’s depreciation expenses in 2024 are expected to exceed NTD 45.3 billion (nearly USD 1.4 billion), and the total depreciation in 2025 may surpass NTD 54.4 billion (nearly USD 1.7 billion), creating operational pressure for the company in the coming year.

On the other hand, VIS announced in June a joint venture with NXP to build a 12-inch fab in Singapore. According to its press release, the construction is set to begin in the second half of 2024, with mass production expected by 2027. The initial investment for the fab is approximately USD 7.8 billion.

As per Commercial Times, VIS’ Singapore factory will begin depreciation in 2026, with the company’s depreciation expenses expected to surge from 2026 to 2027. As the depreciation period for lands, factories, machinery and equipment may last for decades, the report indicates that from 2026 onwards, VIS will also experience operational pressure due to depreciation.

And it is certainly not the only challenge for these foundries to tackle. According to a previous report from the Economic Daily News, prices for 12-inch wafers from Chinese foundries led by SMIC, Hua Hong Group, and Nexchip have been slashed by as much as 40% recently, compared to the offer from Taiwanese companies.

According to TrendForce, Chinese foundries are set to drive the bulk of mature process capacity growth in 2025, thanks to China’s domestic IC substitution policies. It is estimated that the capacity of the world’s top 10 mature process foundries will increase by 6% in 2025, though pricing pressures will persist.

Read more

• [News] SMIC Reports Record Q3 Revenue, Cautious on Expansion Amid Overcapacity Warnings
• [News] SMIC, Hua Hong Reportedly Pressure Taiwan Foundries with up to 40% Discount on Mature Nodes

(Photo credit: UMC)