MediaTek, a leading IC design company, has unveiled its flagship 5G Agentic AI processor, the Dimensity 9400, positioning it as an Arm PC-class processor. The chip features a second-generation all-big-core design, combining Arm v9.2 CPU architecture with advanced GPU and NPU technologies, promising high performance and ultra-energy efficiency. According to a report from TechNews, products powered by the Dimensity 9400 are expected to hit the market soon, setting the stage for a showdown with Apple’s A18 and Qualcomm’s Snapdragon 8 Gen 4.

The Dimensity 9400 employs a second-generation all-big-core CPU, featuring a single Arm Cortex-X925 core clocked at 3.62GHz, three Cortex-X4 cores, and four Cortex-A720 performance cores. Compared to its predecessor, the Dimensity 9300, the new chip offers a 35% boost in single-core performance and a 28% improvement in multi-core performance. Additionally, the Dimensity 9400 is built on TSMC’s advanced 3nm (N3E) process, resulting in a 40% reduction in power consumption, providing users with significantly longer battery life.

The processor also integrates MediaTek’s 8th-gen AI processor, the NPU 890, which supports on-device LoRA training and high-definition image generation, while offering developers access to Agentic AI capabilities. This results in a substantial improvement in AI performance and efficiency compared to the previous generation, with an 80% increase in prompt performance for large language models (LLM) and a 35% reduction in power consumption.

TechNews also highlighted that while the Dimensity 9400 emphasizes its all-big-core design, the configuration differs slightly from the previous Dimensity 9300, which used four Cortex-X4 cores and four Cortex-A720 cores. MediaTek explained that the new design was made for energy efficiency, as users don’t always need to run applications at the highest clock speeds of all big cores. Whether the performance lives up to MediaTek’s claims will be determined once the actual products hit the market.

Even before its launch, the Dimensity 9400 generated significant market interest. Reports suggest that vivo’s X200 series and OPPO’s Find X8 will be among the first devices to feature the new processor. Additionally, Samsung is considering replacing its in-house Exynos processor with the Dimensity 9400 in its upcoming Galaxy S25 series, set for release in 2025.

(Photo credit: MediaTek)

In addition to its latest smartphone chipset Dimensity 9400, MediaTek seems to be working on another killer product. According to the reports by Chinese media outlet IT Home and Tom’s Hardware, the Taiwanese IC design giant is teaming up with NVIDIA on an AI processor, which is expected to be manufactured with 3nm and scheduled to be taped out later this month.

According to the reports, the chip is said to enter mass production by late 2025.

According to Tom’s Hardware, building on insights from Qualcomm’s Snapdragon X Elite launch, there’s room for a new Windows-on-Arm competitor from a mobile-focused company like MediaTek. As Qualcomm’s Snapdragon X Elite is comparatively weak in its graphic performance, this creates a market for NVIDIA and MediaTek to step in.

MediaTek’s upcoming AI processor is reportedly to be paired with NVIDIA’s GPU, while major laptop companies led by Dell, HP, Lenovo and Asus are said to be potential customers for the product, the reports note. According to a previous report by the Economic Daily News, the processor is rumored to be priced around USD 300.

It is also worth noting that MediaTek, which is known for its leading position on mobile chipsets, typically partners with AMD for hardware, with AMD utilizing MediaTek’s Wi-Fi 6E solutions in its mobile platforms, according to Tom’s Hardware. There have also been longstanding rumors of a 5G notebook collaboration between the two companies, the report says.

Therefore, it would be somewhat interesting if MediaTek does turn to NVIDIA, AMD’s biggest rival, for collaboration. However, as NVIDIA currently holds the throne of AI processors, it would not be surprising that MediaTek teams up with the U.S. AI giant when it chooses to expand its foothold to the AI PC sector.

It would not be the first time the two companies team up, though. In March, MediaTek announced at NVIDIA GTC four new automotive SoCs within its Dimensity Auto Cockpit portfolio, offering powerful AI in-cabin experiences for the next generation of intelligent vehicles.

For more details, it integrates an NVIDIA RTX GPU, which supports ray tracing for realistic visuals and lighting effects in games, plus AI upscaling and frame generation for fast, fluid action, according to MediaTek’s press release.

Read more

• [News] MediaTek Ventures into AI Accelerators, Aiming to Release Dimensity 9400 in October
• [News] MediaTek Reportedly Teams up with NVIDIA, Aiming AI PC Chip Launch in 1H25 

(Photo credit:)

There are signs that OpenAI, the company that rose to fame with its AI models, is now eyeing the semiconductor manufacturing sector. However, can building a wafer fab be an easy success?

Recently, international media revealed details of OpenAI CEO Sam Altman’s meetings with senior executives from multiple chip manufacturers during his visit to Asia last year.

Altman visited top executives at companies such as TSMC and Samsung, proposing an ambitious plan to invest $7 trillion to build 36 new wafer fabs and data centers to drive the development of artificial intelligence. Altman envisioned that these fabs, funded by the United Arab Emirates, would produce AI chips, which OpenAI and other companies could use to build AI data centers.

The report highlighted that the scale of the investment Altman mentioned is equivalent to a quarter of the annual output of the U.S. economy. To meet OpenAI’s expansion needs for computing power, it would take several years to complete the necessary wafer fabs.

However, due to cost considerations, TSMC did not endorse Altman’s plan. TSMC executives considered Altman’s proposal too aggressive and risky. Even building a few more wafer fabs involves high risk due to the immense capital required, let alone 36 fabs.

How Much Does a Wafer Fab Cost? Hundreds of Billions of Dollars

In recent years, driven by the demand for AI models, the need for chips has surged, and wafer fabs have been expanding rapidly. However, as OpenAI’s experience shows building a wafer fab is no simple task. It faces challenges such as international dynamics, costs, and technological hurdles, with cost being the largest barrier.

The cost of a wafer fab primarily involves land and facility construction, equipment procurement, technology development and intellectual property, as well as operation and maintenance. Land and facility construction take up a significant portion, as a fab requires extensive land for building plants and basic infrastructure such as electricity, water supply, and communication.

On the equipment side, the purchase of lithography machines, etching machines, ion implanters, and thin-film deposition tools is a major expense, especially for advanced lithography machines, which are extremely costly.

Additionally, a wafer fab requires significant research and operational costs, including intellectual property, equipment maintenance, staff training, safety protocols, and environmental management, all of which demand continuous investment from manufacturers.

When all these factors are calculated, the cost of building a wafer fab is extremely high. Moreover, as chip manufacturing processes evolve, the cost of fabs continues to rise. The industry estimates that the cost of a modern fab is in the range of billions of dollars. For example, Intel’s two factories in Arizona are expected to cost $15 billion each, while Samsung’s fab in Taylor, Texas, is projected to cost $25 billion.

Regional Differences in Wafer Fab Costs

It’s also worth noting that the cost of building a wafer fab varies by region. In Asia, for example, due to a well-established supply chain, abundant talent, and policy support, the cost of building a fab is relatively lower. In regions like Europe, the U.S., and the Middle East, however, costs may be higher due to the need to import technology, train talent, and develop a complete supply chain.

(Photo credit: Intel)

Ahead of MediaTek’s official launch of Dimensity 9400, which takes place on October 9^th, rumors have been circulating that Samsung may feature the chip in its Galaxy S25 next year. Now the first smartphone equipped with the chip has surfaced. According to the reports by Gizmochina, Wccftech and mydrivers, Vivo X200 series will reportedly be the first smartphone powered by the Dimensity 9400 chipset.

It is worth noting that Dimensity 9400 will be reportedly be manufactured with TSMC’s 3nm, according to Wccftech. This will also mark the first 3nm chip in the Android ecosystem, mydriver notes.

For more details, Vivo is anticipated to unveil the Vivo X200 series on October 14th, which includes the Vivo X200, Vivo X200 Pro Mini, and Vivo X200 Pro. According to Gizmochina, all the models will be powered by Dimensity 9400.

Dimensity 9400, by introducing the Arm Cortex-X925 super-large core, is said to offer a 36% performance increase and a 41% improvement in AI performance if compared to Cortex-X4, Gizmochina notes.

While MediaTek’s Dimensity 9400 is said to excel in performance, there have been rumors suggesting that the unit price of the SoC has increased by 20%, which could force smartphone manufacturers to adjust the prices of their flagship models, Wccftech notes.

However, as the start price of X200, which is the base model of Vivo’s latest lineup, has been revealed to be 3,999 Yuan (around USD 570) per unit, the aforementioned concern has been eased, Wccftech indicates.

Recent reports indicate that MediaTek’s Dimensity 9400 is priced 20% lower than Qualcomm’s upcoming Snapdragon 8 Gen 4, which enhance the likelihood for Samsung to adopt the chip in its upcoming S25 to reduce the cost.

Read more

• [News] MediaTek Ventures into AI Accelerators, Aiming to Release Dimensity 9400 in October
• [News] MediaTek Sets Dimensity 9400 Launch Date, Ahead of Qualcomm’s Snapdragon 8 Gen 4

(Photo credit: Vivo)

In an interview with CNBC, Foxconn’s chairman Young Liu pointed out that as advanced language models keep evolving, the investment in AI boom will continue as well. Liu suggested that the AI boom “still has some time to go,” as advanced language models like OpenAI continue to grow increasingly intelligent with each update.

He stated that the current trend in the technology industry is the development of AI that is as intelligent as humans, or even smarter. This type of AI is referred to as “AGI,” or “artificial general intelligence.”

Citing Liu’s remarks, the report noted that among the discussions about AGI and various levels of intelligence, the intelligence can be categorized into four levels. According to Liu, we are currently at level two, with levels three and four yet to be reached. Liu also noted that the advancement of AI towards greater intelligence will undoubtedly benefit the AI server industry and has been a significant driver of Foxconn’s growth this year.

Regarding the demand for NVIDIA’s new generation Blackwell GPU, Liu remarked that the chip demand was “far beyond our expectations.” He also mentioned that Foxconn is constructing a new factory in Mexico to meet this substantial demand. His comments align with those of NVIDIA CEO Jensen Huang, who has previously characterized the demand for Blackwell as “insane,” according to CNBC.

Despite concerns about potential yield issues with NVIDIA’s GB200, the company remains on schedule to deliver the chips by the fourth quarter of 2024, according to Liu. Foxconn, a key contract manufacturer for the GB200, expects substantial growth fueled by strong demand for AI servers.

Furthermore, Liu suggested that generative AI-related devices will be the next growth area, noting that the company is currently witnessing this trend in the cloud, where demand for generative AI cloud devices is exceptionally strong. In the next phase, the market will see further development of generative AI devices, according to Liu.

(Photo credit: Foxconn)