Would Samsung finally be able to secure major orders from tech giants, even one of the world’s top AI companies? According to a report by The Information, the most successful and lucrative partnership in AI business, which is formed by NVIDIA and its foundry partner TSMC, is showing signs of strain, while Samsung may turn out to benefit from the development.

However, it is worth noting that the orders Samsung might get from NVIDIA may not be the most advanced AI chips. According to the reports by The Information and SamMobile, the U.S. AI chip giant is mulling to team up with Samsung to produce its new GPUs, which are considered less complex to manufacture than its AI accelerators.

The reports also suggest that NVIDIA is trying to secure discounted pricing from Samsung, as it aims for a 20-30% reduction compared to what it pays TSMC.

According to the analysis by SamMobile, NVIDIA’s move is an attempt to reduce its dependence on TSMC for upcoming chips, which is a positive development for Samsung.

Though the struggling semiconductor giant is said to have several clients for 5nm, 7nm and 8nm nodes, the continuous yield issues for 3nm and 4nm makes it unable to attract major customers, according to SamMobile, while the company now hopes to improve its yields and regain clients like Qualcomm and NVIDIA.

According to The Information, Samsung’s opportunities arise while NVIDIA’s Blackwell chips, built with TSMC’s 4nm, reportedly faced delays due to issues discovered in the testing process. The chips are said to have failed in high-voltage environments typical of data centers, which put strain on the decades-long partnership between the two firms.

For now, the issues have been resolved, and Team Green’s Blackwell chips are expected to ramp up starting from Q4 2024. According to a report by Wccftech, citing the projection by analyst Ming-Chi Kuo, Blackwell’s estimated shipments are expected to be around 150,000 to 200,000 units this quarter, and surging to 500,000 to 550,000 units in Q1 2025.

According to Kuo, Microsoft is believed to be the major customer, with its orders for GB200 in Q4 skyrocketing, rising 3 to 4 times from the previous range of 300 to 500 racks (primarily NVL36) to about 1,400 to 1,500 racks (approximately 70% NVL72).

Read more

• [News] NVIDIA’s Blackwell Overcomes Delays, as GB200 Reportedly Sets for December Mass Production

(Photo credit: NVIDIA)

Rumors have been circulating recently that NVIDIA, after the shipment of GB200 in the fourth quarter, is mulling to replace the current on-board solution with an independent GPU socket design, which might benefit Taiwanese supply chain companies such as Foxconn and interconnect component supplier LOTES, according to the reports by MoneyDJ and the Economic Daily News.

Industry sources further indicate that NVIDIA is expected to switch from an OAM type (on-board solution) to a socket type soon, starting with the GB300 product line.

It is understood that after the second half of 2025, the B300 series will likely to become the mainstream product for NVIDIA. The main attraction of the B300 series is said to be its adoption of FP4, which is well-suited for inference scenarios.

This change of design is expected to improve the yield rate for GPUs though it might probably reduce performance. According to the Economic Daily News, adopting socket design will help simplify after-sales service and server board maintenance, as well as optimize the yield of computing board manufacturing.

On the other hand, the new design is believed to allow more production flexibility, as manufacturers may not need to be equipped with SMT production line.

It is worth noting that if NVIDIA does initiate the change, this would mark the first time the AI chip giant introduces socket design in its GPU products, the Economic Daily News notes. However, this is not the first instance for the design to be introduced in the industry, as NVIDIA’s archrival AMD has adopted the socket design in 2023 with MI300A, according to the report.

Despite recent speculations concerning Blackwell’s yield rate, a previous report by Commercial Times noted that NVIDIA’s updated version of B200 is expected to be completed by late October, allowing the GB200 to enter mass production in December. In an interview with CNBC, NVIDIA CEO Jensen Huang stated that the demand for the company’s next-generation Blackwell AI chip is “insane,” and that “everybody wants to have the most and everybody wants to be first.”

Read more

• [News] NVIDIA’s Blackwell Overcomes Delays, as GB200 Reportedly Sets for December Mass Production
• [News] Foxconn Remains Unfazed Despite NVIDIA’s GB200 Delay, with Customers Aim to Secure Their Orders

(Photo credit: NVIDIA)

Amid concerns on the delay of NVIDIA’s Blackwell, CEO Jensen Huang spoke at the Goldman Sachs Communacopia and Technology Conference a few hours ago, trying to ease the doubts of the market by expressing his optimism on the demand of its products and the company’s future prospects.

“We have a lot of people on our shoulders, and everybody is counting on us,” said Huang, according to a report by Yahoo Finance. He even joked that as the world relies so much on the company, the engineers may have to spend more time on work. “Less sleep is fine. Three solid hours is all we need.”

Huang also elaborated on the demand for the delivery of NVIDIA’s components, technology, infrastructure and software, stating that it is so overwhelming that people may get “emotional,” as it has a direct impact on their revenue and competitiveness.

It is worth noting that Huang also mentioned that NVIDIA heavily relies on TSMC for producing its most important chips, as in many ways, according to a report by Economic Daily News. He said TSMC’s agility and ability to respond to demand are incredible.

Huang stated that most of NVIDIA’s technology is self-developed, and if necessary, orders could be shifted to other suppliers. However, such adjustments could lead to a decline in chip quality, according to the report.

According to a previous report from Commercial Times, NVIDIA has reportedly executed changes to the Blackwell series’ GPU mask. Therefore, the process can now proceed without re-taping out, with NVIDIA’s updated version of B200 expected to be completed by late October, allowing the GB200 to enter mass production in December.

Moreover, in his latest meeting with Goldman Sachs, Huang noted that the first trillion dollars of data centers is going to get accelerated, creating a new type of software, generative AI.

Citing Huang’s remarks, the report by Yahoo Finance stated that it matters a lot because generative AI is not just a tool but a “skill,” so for the first time, the AI chip giant is developing skills that will enhance human capabilities.

According to Yahoo Finance, Huang said that NVIDIA, along with cloud service providers (CSPs), build the infrastructure in the cloud so developers can access these machines to train, fine-tune, and safeguard models.

It is worth noting that Huang tried to materialize the benefit, saying that for every dollar a CSP spends with NVIDIA, it results in USD 5 worth of rentals. He also said while training AI models is resource-intensive, it pays off in the long run.

Citing Huang, the report stated that NVIDIA’s servers may seem expensive at first glance, as it potentially costs a couple of million dollars per rack. However, they replace thousands of nodes. What is remarkable is that the cost of cables for old, general-purpose computing systems is higher than consolidating everything into a single, dense rack, Huang said.

According to Yahoo Finance, Huang also noted that the days of software engineers writing every line of code are completely behind. In his vision, every software engineer will have digital companions working alongside them 24/7.

In addition, NVIDIA, with its 32,000 employees, hopes to be supported by “100 times more digital engineers” in the near future, the report noted.

Notably, there seems to be another piece of good news for the U.S. chip giant. According to a report by Reuters, the U.S. government is said to be mulling, allowing NVIDIA to export advanced chips to Saudi Arabia, which would enable the country to train and operate the most powerful AI models.

According to the report, Saudi Arabia expects to receive shipments of NVIDIA’s most advanced chips, the H200s, which were first used in OpenAI’s GPT-4o.

Read more

• [News] NVIDIA’s Blackwell Overcomes Delays, as GB200 Reportedly Sets for December Mass Production
• [News] Foxconn Remains Unfazed Despite NVIDIA’s GB200 Delay, with Customers Aim to Secure Their Orders

(Photo credit: NVIDIA)

According to a report from The Register, DPU developer Xockets recently filed a lawsuit, accusing AI chip giant NVIDIA, Microsoft, and intellectual property risk management company RPX of colluding to avoid paying Xockets the fees it is owed, violating federal antitrust laws, and intentionally infringing on its patents.

The report states that in addition to seeking monetary compensation, Xockets is also requesting an injunction. If granted, this injunction would prevent NVIDIA from selling its upcoming Blackwell architecture GPUs.

Per Reuter’s report, Xockets, founded in 2012, claims that its invention, the Data Processing Unit (DPU), plays a critical role in some of NVIDIA’s and Microsoft’s systems. The company states that its technology helps offload and accelerate tasks that would otherwise place a heavy burden on server processors.

Reportedly, Xockets founder Parin Dalal began filing a series of DPU technology patents in 2012. These patents describe architectures used for the linear downloading, acceleration, and isolation of data-intensive computational operations from server processors.

Xockets claims that its DPU-related patents cover various applications including cloud computing, machine learning, security, network overlay, stream data processing, and cloud computing architectures. Xockets alleges that Microsoft and Mellanox, which was acquired by NVIDIA in 2020, which was acquired by NVIDIA in 2020, have infringed on these patents.

In a recent statement, Xockets claimed that NVIDIA has utilized DPU technology patented by Xockets, allowing NVIDIA to monopolize the AI server market using its GPUs. Meanwhile, Microsoft has allegedly monopolized the AI platform market using NVIDIA GPUs.

Xockets further claimed that it has made effort to engage in sincere negotiations with NVIDIA and Microsoft, but these attempts have been rejected.

Xockets’ lawsuit reveals that it actually demonstrated the relevant technology to Microsoft in 2016, and the technology was subsequently adopted by Mellanox within the same year for cloud computing downloads used by Redmond and other clients.

Additionally, NVIDIA’s ConnectX smartNIC, BlueField DPU, and NVLink switch, which are crucial for extending AI training and inference deployments across large GPU clusters, are said to infringe on Xockets’ patents.

Regarding this matter, NVIDIA has declined to comment, while Xockets’ spokesperson has also not provided any additional explanation.

The report highlights that Microsoft and NVIDIA may not be Xockets’ only targets but are at least the most profitable ones. Other companies, such as Broadcom, Intel, AMD, Marvell, Napatech, and Amazon, are also actively developing products similar to NVIDIA’s ConnectX, BlueField, and NVLink.

Regarding the lawsuit, the judge overseeing the case has approved a preliminary injunction hearing to be held on September 19.

Read more

• [News] NVIDIA’s Blackwell Overcomes Delays, as GB200 Reportedly Sets for December Mass Production
• [News] Foxconn Remains Unfazed Despite NVIDIA’s GB200 Delay, with Customers Aim to Secure Their Orders

(Photo credit: Xockets)

According to a report from Commercial Times citing sources, it’s revealed that NVIDIA has executed changes to the Blackwell series’ 6-layer GPU mask. Therefore, the process can now proceed without re-taping out, as production delays being minimized.

The report noted that NVIDIA’s updated version of B200 is expected to be completed by late October, allowing the GB200 to enter mass production in December, with large-scale deliveries to ODMs expected in the first quarter of next year.

Previously, as per a report from The Information, NVIDIA’s GB200 was said to be experiencing a one-quarter delay in mass shipments. Another report from the Economic Daily News further suggested that the problem likely lies in the yield rates of advanced packaging, which mainly affected the non-reference-designed GB200 chips.

Industry sources cited by Commercial Times addressed that NVIDIA’s Blackwell chip used to be facing instability in metal layers during the HV process, which was then resolved by July.

In addition, since the issue reportedly occurred in the back-end-of-line process, a new tape-out was deemed unnecessary. Still, as CoWoS-L capacity remains a bottleneck, the advanced packaging for GB200 this year is expected to adopt  CoWoS-S.

Read more

• [News] Foxconn Confirms Shipment of NVIDIA’s GB200 to Start in Q4 as Planned
• [News] Foxconn Remains Unfazed Despite NVIDIA’s GB200 Delay, with Customers Aim to Secure Their Orders

(Photo credit: NVIDIA)