Samsung Memory Business’ Worst Year Since 2009

For context, 2023 wasn’t the best year for Samsung’s DRAM business. The chipmaker’s largest profit-generating segment, semiconductors, was under pressure for most of the year—losses of over 4.58 trillion, 4.36 trillion, and 3 trillion won were recorded in the first three quarters of the year, respectively. The fourth quarter was the least damaging, with ~ only 14% losses recorded. These were the worst numbers that the company had seen since 2009.

While Samsung saw a mind-boggling 95% decline in profitability in 2023, the statistics above and the decreasing loss pattern tend to provide a solid basis for understanding the general trend—recovery is slow but imminent.

Samsung is the largest memory chipmaker in the world, but in a volatile industry, the company needs to keep investing. Looking at the past 15 months, this has come off as draining money into a pithole. Still, according to research firm Omdia, Samsung may end up finally recovering to similar levels to last year in the second half of 2024.

Memory Semiconductor’s Recovery Trend; Might Recoup Profits in H2 2024

Samsung’s wafer production numbers for the first half of 2024 accumulated to about 1.575 million units, 75% of the 2.1 million units it produced in the same quarter last year. However, this 25% gap is expected to close to about 0% as the predicted forecast numbers are 1.785 million units, the same as those of Q2 2023.

This upward trend will likely not stop here, as production will ramp up to 1.965 million units in Q3 2024, this time overtaking the production numbers of the third quarter of last year, which were 1.77 million units, marking an 11% improvement. For the last three months of 2024, the quarterly projection is expected to cross the 2 million barrier, this time marking a 41% rise from that of Q4 2023, which came at about 1.425 million units.

HBM & AI Chips Will Be the Main Profit Drivers

Samsung will now be looking to recover lost profits. It will do this by selectively focusing on its memory business. For example, it will pour more resources into its HBM chips, which will, in turn, improve profit margins. This is why a report from SeDaily points out that production will increase, but only along specific lines.

The Hwaseong 15, Pyeongtaek facility, which produces HBM products, is seeing increased production numbers, but others, like the Hwaseong 13 line, which makes image sensors, are still looking at numbers worse than that of the same period last year.

This is all we know for now, but rest assured that we will keep you updated as new information becomes available.

The post Samsung Memory Chip Business is Expected to Make a Comeback This Year, Profit Recovery Forecasted in H2 2024 appeared first on Appuals.

Apple’s A18 Pro May Miss Out on the LPDDR6 Upgrade this Year

Apple Silicon’s next generation of flagship mobile APs, presumably titled A18 Pro (to be used in iPhone 16 Pros), may skip this upgrade and stick to LPDDR5T. According to Korean news outlets, the Snapdragon 8 Gen 4 may be the first mobile SoC ever to support LPDDR6 in commercial smartphones.

With these SoCs, the main hurdle lies in the foundries’ partners, as they are the ones who make the final call on sourcing the right set of memory modules that suit their plans. This is why Vivo’s X100 Pro, which was released earlier last November, was rumored to come with LPDDR5T, but it, too, skipped the upgrade and settled with LPDDR5X.

Since Apple may not incorporate LPDDR6, Qualcomm’s 8 Gen 4 may add another advantage to its tally. This time, the Android SoC manufacturer may keep it close to the industry leader, Apple, which has been at the top for almost 11 consecutive years.

The LPDDR6 DRAM Modules Will Be Finalized By the 2nd Half of 2024

Samsung and Sk Hynix, two of the world’s leading DRAM manufacturers, are ramping up the production of these LPDDR6 modules. According to Ajunews, the International Semiconductor Standards Organization (JEDEC) will finalize these modules by the second half of this year, so it is safe to assume that the chances of it being incorporated into Snapdragon 8 Gen 4 are slim.

The same source also pointed out that LPDDR7 may debut in NVIDIA’s upcoming set of RTX 5000 series and AMD’s RDNA4 GPUs.

This is all we know for now, but rest assured that we will keep you updated as new information becomes available.

via: Ajunews

The post Snapdragon 8 Gen 4 May Push for LPDDR6 Support This Year, Apple A18 to Miss Out appeared first on Appuals.

Downswing in “Annual” Revenue for the World’s Leading Foundries

While the fourth quarter of 2023 did show upward trends in the revenue generated by the top 10 foundries, the YoY results unsurprisingly deviated from this pattern. The annual results showed a 13.6% decline in revenue, totaling $111.54 billion.

Analysts at TrendForce have pointed out that the reasons for this downward spiral go beyond the slowing economy. For one, the foundries had to deal with full inventories across the supply chain. Also, one of the largest consumer markets in the world, China, saw a slow recovery.

AI to Pull Back Revenue Numbers This Year

In 2024, it is predicted that the industry will see a massive rise, up to about 12%, mainly thanks to the rise of AI applications worldwide. This would then place the cumulative revenue for the Top 10 foundries at around $125.24 billion.

TSMC Widens the Gap Between #2, IFS Pushed Back Out of the Top 10

These recent findings show that the Top 10 foundries in the world capture 88.8% of the market, and TSMC alone gets more than 60%, making it the undisputed leader in the industry. IFS, which recently claimed to overthrow TSMC as the world’s leading foundry, has now dropped out of the Top 10.

IFS, which recently announced its 14A process node and 18A CPUs, has seen a decline in revenue in the last quarter of 2024. This is mainly attributed to the shift in older and newer processes, but with the company’s own projections, they aim to climb up to at least the Top 5 in the next six years.

For TSMC, shipments improved by 14% to $19.66 billion, and revenue generated from processes thinner than 7nm increased from 59% to 67%. Again, with increasing orders and the rise of AI applications this year, TSMC is expected to be the primary beneficiary of this change in trend.

China’s SMIC saw a quarterly increase of 3.6%, with revenue coming in at $1.68 billion. Also, with IFS out of the Top 10, Nexchip climbs back into the rankings.

This is all we know for now, but rest assured that we will keep you updated as new information becomes available.

via: TrendForce

The post Intel Foundry Drops Out of Top 10 as Leading Foundries See Strong Growth in Quarterly Revenue appeared first on Appuals.

Companies are Looking for Other Options in the Mobile AP Segment, but It Probably Won’t Affect MTK or QC’s Standing

Part of the reason we have been seeing a steady decline in the performance metrics of mid-range smartphones and an increase in prices for flagships is merely these hikes. This is why companies like Samsung prefer to stick to Exynos. Revegnus on X also points out that “I believe Samsung raised the price of the S24 Ultra this time to minimize the impact of future Qualcomm chip price increases and to prepare in advance.“

Competition is rising in the semiconductor sector, which is mainly fueled by the rise of AI applications. However, both of these have a considerable standing in the mobile application chip market. MediaTek, for instance, accounted for 33%, and Qualcomm stands second with 28%, as of Q3 2023 [Counterpoint, 2024].

The Reasons For These Price Hikes Aren’t Just Limited to Increased Profit Margins

Analysts have predicted various reasons why chip designers are increasing their prices. For one, the world’s largest chipmaker, TSMC, has seen an increase of 30% in electricity billing from the Taiwanese government. On the other hand, Qualcomm is also spending a lot of resources on its in-house Nuvia-based chip solutions. Not to mention how Nuvia itself was acquired by Qualcomm not long ago for a whopping $1.4 billion.

Moreover, as foundries continue to fab thinner nodes, they will eventually need to hike prices to stay competitive; otherwise, they’ll simply fall behind the competition. Qualcomm and MediaTek, both in particular, have been following the industry leader, Apple Silicon, in fabricating the most efficient and best-performing chips in the mobile segment.

It remains to be seen how companies react moving forward. Most might look towards Exynos for its mid-range offerings. Still, others, like Samsung, which are more prominent clients and have struck out a multi-year deal, may eventually squeeze out a favorable deal, but again, it is highly unlikely that we’ll see the same price for flagship phones as we used to before one or two years.

This is all we know for now, but rest assured that we will keep you updated as new information becomes available.

via: Revegnus

The post Qualcomm and MediaTek’s Chip Price Hike Could Drive Mobile Companies Towards Exynos SoCs appeared first on Appuals.

Earlier this year, on January 19th, Microsoft published a blog post revealing that the hacking group Midnight Blizzard (Nobelium), backed by the Russian government, had breached one of the company’s legacy internal testing accounts.

Midnight Blizzard Reportedly Brute-Forced its Way into One of the Company’s Legacy Test Account

Microsoft claims that its internal systems were not found to be vulnerable, but the breach resulted from the hackers using a “spray attack” technique, essentially brute-forcing their way through a list of usernames and using an array of possible password combinations with them. Since the breached account did not have 2FA enabled, the hackers were able to easily gain access.

Once Nobelium gained access to one of Microsoft’s accounts, it used its permissions to attempt to infiltrate corporate accounts, specifically targeting Microsoft’s senior leadership. This was allegedly because “they were initially targeting email accounts for information related to Midnight Blizzard itself.“

Now, the group has shared some of its findings and secrets with Microsoft itself and its customers. According to Microsoft, Midnight Blizzard, or Nobelium, has also increased the frequency of its spray attacks by as much as tenfold this month compared to January 2024.

Microsoft has recently fallen victim to numerous cyberattacks. Just three years ago, in April 2021, several US organizations’ emails were hacked as part of a Microsoft Cloud exploit. This year as well, the Nobelium attacks have been reported for over four months now, and Microsoft warns that the hackers will use the obtained information “to accumulate a picture of areas to attack and enhance its ability to do so.“

This is all we know for now, but rest assured that we will keep you updated as new information becomes available.

via: Microsoft

The post Microsoft Claims Russian Hackers Gained Access to its Source Code via Email Breach appeared first on Appuals.

Meta experienced a massive outage late on Tuesday, affecting all of its platforms, namely Facebook, Instagram, and Messenger. Users worldwide who are attempting to access their accounts have been logged out of the service.

Downdetector Sees a Spike in Outage Reports Concerning Meta Platforms

Downdetector, a real-time outage monitoring service, has reported a surge in outage reports around 8:20 PM GMT+5. Currently, it shows over 200,000 reports. Users who are trying to log back in are coming across an “unexpected error occurred” message.

At the time of writing, Meta has not issued a public statement regarding the situation. However, given the rare nature of the outage, it’s understood that the team is likely working to fix the issue as soon as possible.

Users are also Reporting Issues with WhatsApp

Alongside Facebook, Instagram, and Messenger, some users have also noticed WhatsApp not working. However, the issue with WhatsApp is limited to specific regions only.

The last instance of such an outage was seen back in 2021 when Meta’s platforms were down for nearly seven hours. As of now, reports are coming in indicating that the situation is improving in some regions. Therefore, it is only a matter of time before the service is fully restored.”

Elon Musk Mocks Meta Outage on X

Facebook and Instagram’s rival platform, X apparently saw the opportunity and joked about the whole situation when CEO, Elon Musk tweeted “If you’re reading this post, it’s because our servers are working,” which the official X account retweeted captioning “we know why you’re all here rn.“

This is all we know for now, but rest assured that we will keep you updated as new information becomes available.

The post Facebook, Instagram and Messenger are Down Worldwide, Thousands Suddenly Logged Out appeared first on Appuals.

Despite retaining the services of an attorney, Tropic Haze has come forward in favor of Nintendo saying, “Yuzu is primarily designed to circumvent and play Nintendo Switch games.” Additionally, the company also agrees that it will permanently part ways with Yuzu in terms of working, hosting, distributing code or features, promotion, or doing anything that violates Nintendo’s copyright protection.

If that wasn’t enough, the company will also give in the yuzu-emu.org domain name to Nintendo, delete all copies of Yuzu or circumvention tools used for developing it such as Hekate, TegraRcmGUI, Atmosphere, Lockpick_RCM, NDDumpTool, nxDumpFuse, and Tegra Explorer.

Apart from the digital tools, Nintendo has also asked to hand over physical circumvention devices and modified Nintendo hardware of any kind. There’s a lot more to it and if you want you can read the entire proposed final judgment and permanent injunction, but they are still waiting for approval from a judge.

We still don’t know whether Yuzu is actually circumventing Nintendo’s protections since it doesn’t contain Nintendo’s keys and is a “bring-your-own-BIOS” emulator. However, Nintendo and Tropic Haze are asking a federal judge to acknowledge that Yuzu’s functionality involves circumventing Nintendo’s copyright safeguards, regardless of whether the encryption keys are used with the software or not.

Developing or distributing software, including Yuzu, that in its ordinary course functions only when cryptographic keys are integrated without authorization, violates the Digital Millennium Copyright Act’s prohibition on trafficking in devices that circumvent effective technological measures, because the software is primarily designed for the purpose of circumventing technological measures.

As Yuzu is an open-source emulator, it’s not easy to put a nail in its coffin as copies of both the emulator and its source code are still out there. Some users have also mentioned backing up the code after hearing about the legal case two weeks ago.

That’s all we have for now, let us know your thoughts below about the future of Yuzu and whether this result could impact other emulators!

Yuzu is Shutting Down

Bunnei from the Yuzu team has now confirmed that both Nintendo Switch and Nintendo 3DS emulators will be shutting down; “Yuzu and Yuzu’s support of Citra are being discontinued, effective immediately.”

While Yuzu never intended to promote piracy in any way, their projects can circumvent Nintendo’s copyright protection measures and allow users to play games outside of authorized hardware, leading to extensive piracy. Yuzu has also said that they will be shutting down their code repositories and will be discontinuing their Patreon accounts, Discord servers, and website.

The post Yuzu Lost Against Nintendo And Will Have To Pay $2.4M appeared first on Appuals.

Changing The AI Game With Claude 3

While Claude 3 has three different models, the star of the lineup Opus is more capable than any other AI system currently on the market. According to Anthropic cofounder and CEO Dario Amodei; “Opus is capable of the widest range of tasks and performs them exceptionally well.”

On top of that Amodei also explains how Opus outstrips top AI models like GPT-4, GPT-3.5, and Gemini Ultra on academic benchmarks like GSM-8k for basic mathematics, MMLU for undergraduate-level expert knowledge and GPQA for graduate-level expert reasoning. Amodei says “It seems to outperform everyone and get scores that we haven’t seen before on some tasks.”

Despite not fully disclosing how powerful their AI model Opus is, the reported benchmarks are a clear giveaway that Opus is head-to-head or even above major alternatives like GPT-4 and Gemini Ultra in complex tasks that require advanced reasoning. This establishes Opus at least on paper as a new high standard for commercially accessible conversational AI.

Rest of the Lineup

Apart from Opus, Claude 3’s Sonnet is the mid-range model that offers businesses a cost-effective option to perform routine tasks like data analysis, knowledge work, and more. Meanwhile, Haiku is designed to be fast and affordable, ideal for uses like customer chatbots where speed and cost matter most. Although Haiku is not publicly out yet, the startups expect it to launch “in a matter of weeks, not months.“

The Visual Improvements

The Claude 3 AI models are not only efficient in text-based queries but also showcase extraordinary computer vision capabilities on par with other models. This gives users the option to extract text from images, documents, charts, diagrams, and a lot more.

A lot of [customer] data is either highly unstructured or in some sort of visual format. Just the process of having to manually copy that information to even be able to have it interact with a generative AI tool is quite cumbersome.

Daniela Amodei

The Future of AI

Anthropic’s launch of the Claude 3 lineup is a testimonial to how companies are pushing boundaries to introduce models like Opus capable of outperforming big names like GPT-4 and Gemini Ultra. This is all we have for now, let us know your thoughts on Anthropic’s Claude 3 in the comments down below!

The post Antrhopic’s Claude 3 Outperforms GPT-4 and Gemini Ultra in Benchmark Tests appeared first on Appuals.

SF to Use NVIDIA’s Tech to Tone Down Defective Production Numbers

IFS aims to increase its production volumes and capture the second position in the foundry business. To counter this threat, Samsung is planning a rather unique solution to address its low-yield situation more cautiously. The company reportedly plans to implement NVIDIA’s “Digital Twin” technology in its factories starting next year.

NVIDIA’s Digital Twins (Based on Omniverse Replicator)

Digital Twin technology, like the name suggests creates a controlled and safe virtual space within which a virtual simulation of real-world situations and scenarios for a factory can be run.

A digital twin is a virtual representation — a true-to-reality simulation of physics and materials — of a real-world physical asset or system, which is continuously updated.

What is a Digital Twin — NVIDIA

The simulation uses AI and big data to analyze the production process, searching for trends and variations. This flexibility also allows engineers to experiment with different methods with minimal resource loss, leading to better optimization and efficiency within the whole process, potentially improving yields.

Samsung to Present its Plans for Digital Twin Semiconductor Factories at NVIDIA’s AI Conference this Year

The Digital Twin technology is built on NVIDIA’s Omniverse platform. Samsung will announce its plans for the first Digital Twin-based semiconductor factory at the upcoming NVIDIA GTC 2024 conference, held from March 18-21. Yun Seok-jin, Executive Vice President of Samsung Electronics DS Division Innovation Center, will present the company’s plans during the event.

This initiative is not new. In fact, Samsung established the Digital Twin Task Force in 2022 and subsequently appointed Lee Young-woong, a field expert, as the task force leader. If everything proceeds according to plan, Samsung could potentially begin pilot operations as early as this year, and develop the world’s first Digital Twin Level 5 factory within two years. For reference, this is the highest level among smart factories.

Digital Twin will surely allow for defective product prevention and product and production analysis via AI, but to still counteract the threat of IFS and TSMC stomping over its position in the market, it’ll still have to satisfy clients and solidify customer relationships.

This is all we know for now, but rest assured that we will keep you updated as new information becomes available.

The post Samsung Foundry Turns to NVIDIA’s “Digital Twin” Technology to Counter Yield Issues and Competition from TSMC and IFS appeared first on Appuals.

Musk, the Co-Founder of OpenAI wanted the Artificial Intelligence Firm to Keep a Non-Profit Status

For those who don’t know, Elon Musk was one of the co-founders of ChatGPT-maker, OpenAI back in 2015 when the company signed an agreement to create open-source artificial intelligence systems to benefit humanity as a whole. Three years later, Musk stepped down from the board in 2018 because he wanted to focus on other ventures.

However, Musk now claims that OpenAI’s recent partnerships, such as the one with Microsoft, have diverted the organization from its initial mission. According to the lawsuit, Musk alleges that Brockman and Altman convinced him to invest in the company under the pretense of eliminating the competitive threat posed by Google.

The ‘For-Profit’ Model Following ChatGPT’s Launch

Following the success of ChatGPT, Musk further alleges that OpenAI has embraced a for-profit model, prioritizing the commercialization of their research alongside Microsoft, a global technology giant, that has invested over $13 billion to date.

The lawsuit clearly highlights Musk’s dissatisfaction with OpenAI’s recent shift in priorities, which is particularly concerning given his significant financial contributions as the company’s largest investor from 2016 to 2020, exceeding $44 million. Not only that, but he also claims to have been offered a stake in the company which he ultimately declined due to ethical concerns.

This is all we know for now, but rest assured that we will keep you updated as new information becomes available.

The post OpenAI Co-founder, Elon Musk Sues Altman Over Contract Breach, Accuses the Firm of Going “For-Profit” appeared first on Appuals.