Samsung Electronics on Tuesday projected a far worse than expected 56% plunge in Q2 2025 Operating Profit, claiming that this is  due to weak AI chip sales, deepening investor concerns over the tech giant's ability to revive its struggling semiconductor business.

Q2 2025 Operating Profit Samsung projected an operating profit of 4.6 trillion won for Q2 of 2025, reflecting a 56% drop compared to prior estimates.  Previous Quarter Performance In the Q2 of 2025, Samsung reported an operating profit of 6.7 trillion won. A very big drop.  A more detailed examination of the factors that contributed to Samsung's profit decline in 2025:

1. Chip Industry Challenges:

Oversupply: The semiconductor market faced a significant oversupply situation, particularly in the memory chip sector. This oversupply arose from previous overproduction during a period of high demand, leading to excess inventory. As a result, prices for memory chips plummeted, severely impacting profit margins for companies like Samsung that rely heavily on this segment for revenue.

Weak Demand: There was a notable decline in demand for memory chips, especially from key sectors such as data centers and consumer electronics. The slowdown in the global economy led to reduced investment in data infrastructure and hesitance among consumers to purchase new electronics, further exacerbating the situation.

2. Increased Competition

Rival Companies: Samsung faced intensified competition from other semiconductor manufacturers, notably SK Hynix and Micron. These competitors not only matched Samsung's production capabilities but also engaged in aggressive pricing strategies to capture market share. This competitive pressure forced Samsung to lower its prices, which negatively affected its profitability.

Emerging Players: The rise of new entrants in the semiconductor market also contributed to increased competition. These companies often offered innovative products at lower prices, compelling established players like Samsung to rethink their pricing and marketing strategies.

3. Economic Conditions
Global Economic Slowdown: The global economy experienced a slowdown, which negatively impacted consumer spending across various sectors. Economic uncertainties led consumers to prioritize essential purchases over electronics, causing a decline in sales for Samsung’s products.

Inflation and Interest Rates: Rising inflation rates and increased interest rates affected consumer purchasing power and business investments. Higher costs of living meant that consumers had less disposable income to spend on non-essential items, including smartphones, televisions, and other electronic devices.

4. Investment in AI

High R&D Costs: Samsung made substantial investments in artificial intelligence (AI) and related technologies as part of its long-term strategy to remain competitive in the rapidly evolving tech landscape. While these investments are crucial for future growth, they resulted in high research and development (R&D) costs that strained short-term profitability.

Transition Period: As Samsung transitioned to incorporating AI into its products and services, there was a lag in the realization of returns on these investments. This transitional phase often leads to increased expenses without immediate revenue benefits, contributing to the overall profit decline.

5. Supply Chain Disruptions

Logistical Issues: Ongoing supply chain disruptions, exacerbated by geopolitical tensions and the lingering effects of the COVID-19 pandemic, affected Samsung's production capabilities. Delays in sourcing raw materials and components led to inefficiencies in manufacturing processes.

Increased Costs: The disruptions not only delayed production but also resulted in increased operational costs. Samsung had to navigate higher shipping costs and longer lead times, which further squeezed profit margins.

Conclusion - The combination of these factors created a challenging environment for Samsung in 2025, leading to a significant decline in operating profit. The company’s reliance on the semiconductor market, coupled with external economic pressures and strategic investments in future technologies, highlighted the complexities of navigating a rapidly changing industry landscape. As Samsung looks to rebound, addressing these challenges while leveraging its strengths will be crucial for restoring profitability and sustaining growth

A U.S. judge on Tuesday rejected Huawei Technologies' bid to dismiss most of a federal indictment accusing the Chinese telecommunications company of trying to steal technology secrets from U.S. rivals, and misleading banks about its work in Iran. In the 16-count indictment that Huawei engaged in racketeering to expand its brand, stealing trade secrets from 6 companies, and committing bank fraud.  The Iran accusations stemmed from Huawei's alleged control of Skycom, a Hong Kong company that did business in Iran.
Donnelly said prosecutors satisfactorily alleged Skycom "operated as Huawei's Iranian subsidiary and ultimately stood to benefit, in a roundabout way," from +$100 million of money transfers through the U.S. financial system. The criminal case began during U.S. President Donald Trump's first term in 2018, the same year the Department of Justice launched its China Initiative to address Beijing's alleged theft of intellectual property.

Huawei hoped to wean China off NVIDIA's dominant GPUs via Ascend 910C GPUs, but Huawei continues to encounter substantial resistance to this, courtesy of NVIDIA's ecosystem lock-ins via the CUDA software and Huawei's own shortcomings. It's now reported that China's tech titans, including TikTok's parent ByteDance, Alibaba, and Tencent, have yet to order Huawei's AI chips in large quantities.  Several factors create a sizable inertia around Huawei's 910C GPUs: 

(1) Many of China's tech titans are invested in NVIDIA's CUDA ecosystem, and any break from NVIDIA will entail a significant investment of time and resources, many expect Huawei to adapt to their platforms instead of the other way around.

(2) China's largest tech companies are competitors of Huawei and, feel reluctant to go all-in on their competitor's offering.

(3) Huawei's Ascend 910C ICs suffer from periodic overheating issues, that affects the perception of reliability.

(4) China's largest tech companies stashed away a sizable repository of NVIDIA's GPUs over the years, inventory has yet to run out

(5) US Department of Commerce issued sweeping guidance that any company that using Huawei's ICs without prior authorization could be deemed to be in violation of US export controls.

South Korea's newly elected President Lee Jae-myung has unveiled a flagship policy offering up to 10% production tax credits for semiconductors manufactured and sold domestically, marking a shift from the country's traditional investment-based incentives to output-focused support similar to US measures under the Inflation Reduction Act.

Shanghai-based UniVista Industrial Software Group recently announced it now offers Chinese users free access to Electronic Design Automation (EDA) software. UniVista called the announcement a “technological counterattack” indicating it is responding to blockades and restrictions imposed by the U.S.. Reuters reported that the Trump administration has told US-based companies to stop selling software used to design chips to Chinese customers. Leading companies like Cadence, Synopsys, and Siemens EDA were notified. The 3 companies account for 70% of China’s EDA market, it was reported that Synopsys has told its staff in China to halt sales and stop taking new orders. UniVista, founded in 2020, comprises multiple former Execs of Synopsys and Cadence, several of whom held Senior VP and Fellow-level positions. Besides UniVista, prominent players in China’s EDA ecosystem include Empyrean Technology, Primarius Technologies and Semitronix. As of 2024, Empyrean held 6% of China’s market, Chiana's largest EDA player. EDA tools are software for assisting engineers in designing, simulating, and verifying ICs, PCBs, and SoCs. EDA tools are considered essential for IC design processes. UniVista’s offers EDA tools, including a simulator debugger, a DFT platform, and a PCB design platform.

North Korean hackers operated a “laptop farm” scheme using fake identities to land remote US tech jobs and illegally collect $17.1 million in wages. Part of a broader effort to exploit global labor markets through cybercrime. Cybersecurity experts described the operation as “something we’d never seen before,” citing sophisticated tactics and custom-built programs enabling North Korean agents to bypass detection systems and exfiltrate sensitive corporate data. Federal court documents and a report from the WSJ, North Korean IT workers ran a covert operation where dozens of US-based laptops, remotely controlled using compromised identities, allowing them to pose as job seekers in the US tech sector, securing employment with US companies. Once hired, the “workers” funneled salaries, using proxy accounts and intermediaries. Several Americans unknowingly or willfully participated by setting up and maintaining laptop farms, receiving employer-issued hardware, and managing employment documents, giving North Korean operatives direct access to corporate systems.  North Korea’s strategy of infiltrating remote jobs is not limited to the US; the UK and EU countries were as well. More aggressive tactics are being deployed in these regions, with the hackers threatening to leak proprietary information if their contracts are terminated. This evolving pattern highlights North Korea’s ability to adapt its cyber tactics across borders, turning the global remote work economy into a new frontier for illicit revenue and intelligence gathering.

Intel's Private Equity CEO Lip-Bu Tan is shaking up the company’s organizational structure and cutting costs but isn’t remaking its strategy. “I think it’s a fair assessment that Lip-Bu isn’t thinking about massive changes,” CFO David Zinsner told a J.P. Morgan conference in Boston. Intel spent months in limbo after its board forced out ex-CEO Pat Gelsinger in early December. Company executives openly discussed the possibility of breaking up the business, separating its chip design business from Intel’s manufacturing arm.  Once the world’s largest chipmaker, Intel is trying to recover from an extended decline in sales and a $19 billion loss in 2024. A plan pushed by Gelsinger, and only later embraced by Tan, aims to open Intel’s factories to make chips for other companies.  Since Intel hired Tan in March, he's emphasized taking steps to improve Intel’s decision-making and has said nothing about splitting the business. Instead, Tan has said he plans to reduce the layers of management and implement “several months” of job cuts to cut costs and flatten the organization. “When he looks at the business, what he feels is the biggest issue at this point is a lack of execution”.  Intel hasn’t said how many jobs it will cut altogether. But telling factory workers last week that it is making layoff decisions this month and will notify staff by the middle of June.  Tan said, “We must balance our reductions with the need to retain and recruit key talent. Removing layers of management will give engineers more voice in product design, and a closer connection with Intel customers. He also said fewer layers of management will provide Tan a clearer view of how the company is operating.  “What he wants is the lowest level of the organization to be closer to him,”  Intel has been slow to attract customers to its contract manufacturing business, (Intel Foundry). Intel doesn’t have significant commitments from potential clients yet for its new 18A manufacturing node. Intel still expects its factory business will break even in 2027, Zinsner saying it's “a little bit embarrassing” that Intel hasn’t been delivering better products to customers. Intel cut 15,000 jobs in 2024, and employees say privately that morale is low this spring with the prospect of additional layoffs to come. Some employees are also unhappy with Tan’s decision to order workers back to the office 4 days a week. Zinsner blamed remote work for some of the company’s issues. “We did have a number of the team that was not working on site full time,”  “I think to some degree that has inhibited us and impacted our ability to execute on road maps and process.”  Intel is now walking a narrow path, cutting costs, jobs and employee perks while trying to retain and recruit top technologists to improve its products. Zinsner said Tan is “a magnet for talent” stating it will become apparent soon that Intel is again a destination for top-level professionals in the semiconductor industry.

“We have lost talent at the company, and we do need to rebuild that.”

Huawei cannot legally obtain advanced chips made by TSMC, the company used shell companies last year to obtain compute chiplets for its Ascend 910 AI chips. The conspiracy was discovered by TechInsights and TSMC, which ceased to ship chiplets to Huawei's proxies and initiated an internal investigation. However, it was not clear how many chiplets it supplied to Huawei. According to a report by the Center for Strategic and International Studies, Huawei obtained as many as two million Ascend 910 AI chiplets. "However, TSMC manufactured large quantities of Huawei Ascend 910B chips on behalf of Huawei shell companies and shipped the chips to China in violation of U.S. export controls," the report by CSIS reads.

According to the report, "government officials told CSIS that TSMC manufactured more than 2 million Ascend 910B logic dies and that all of these are now with Huawei. If true, this is enough dies to make 1 million Ascend 910C units. […] Even though Huawei likely has the more than 2 million Ascend 910B logic dies made by TSMC, there is a question as to whether it has enough HBM to integrate with those dies […] It seems likely that Huawei does, however, since the U.S. plan to restrict all advanced HBM sales to China on a country-wide basis was leaked to Bloomberg in August 2024 and did not go into effect until December of that year, giving Huawei ample time to legally acquire HBM chips as part of a stockpiling strategy." Although the report seems correct about Huawei's stockpiling strategy and even gives us an insight into how many chips TSMC produced for Huawei's intermediaries, it still contains several inaccuracies that lead to wrong conclusions.  Huawei's original HiSilicon Ascend 910, which was launched in 2019, consists of a Virtuvian AI chiplet, a Nimbus V3 I/O die, four HBM2E memory stacks, and two dummy dies. TSMC produced Virtuvian chiplets for Huawei from 2019 to September 2020, using its N7+ process technology, a 7nm-class node with some EUV layers.  After the U.S. government put Huawei on its Entity List in 2020, Huawei had to redesign its Virtuvian chiplet to make it at SMIC, which used its N+1 technology (1st Generation 7nm-class process) to build it. GPUs with the new Virtuvian chiplet are called HiSilicon Ascend 910B and have nothing to do with TSMC. Later, Huawei developed a more sophisticated version of its Virtuvian chiplet for its Ascend 910C, which SMIC makes using its 2nd Generation 7nm fabrication technology (N+2). Contrary to the report, the Ascend 910C has only one compute chiplet. Again, the Ascend 910C has nothing to do with TSMC. As Huawei managed to deceive TSMC, the latter produced the original Ascend 910 chiplet for the company in 2023 – 2024, as discovered by TechInsights. Another noteworthy thing about Huawei's Ascend 910B and Ascend 910C is that their yields are not exactly high, so most parts are shipped with some compute elements disabled. Also, only 75% of Huawei's AI chips survive advanced packaging, which is not a good result. "However, the advanced packaging process by which two Ascend 910B dies and HBM are combined into a unified Ascend 910C chip also introduces defects that can compromise the functionality of the chip," the report says. "Industry sources told CSIS that roughly 75% of the Ascend 910Cs currently survive the advanced packaging process." Nonetheless, Huawei continues to acquire millions of Ascend 910B and Ascend 910C for its internal AI projects and external customers. For example, DeepSeek claims that the Ascend 910C delivers 60% of the performance offered by Nvidia's H100, which may not be enough for training large language models but is good enough for inference workloads.

A former SK Hynix employee has been formally accused of illegally transferring technologies related to advanced chip packaging used for 3D NAND, HBM, and multi-chiplet assemblies as well as CMOS image sensors to Huawei's HiSilicon division, reports DigiTimes citing the Seoul Central District Prosecutor's office. An interesting twist to this story: While working at Huawei, the suspect allegedly took IP related to CMOS image sensors (CIS) and hybrid bonding chip packaging technology that is used for 3D NAND, HBM3 / HBM3E / HBM4 memory, as well as Advanced Stacked multi-chiplet packages (e.g., TSMC's SoIC and Intel's Foveros 3D). Huawei did not explicitly ask the suspect to bring it SK hynix's trade secrets (even though the company is known for poaching highly-skilled employees from world-class companies by offering them huge salaries), but accepted them as part of the hiring process.   It is unclear whether the suspect's targets were hybrid bonding and image sensors, or 'just' hybrid bonding as modern CMOS sensors also use hybrid bonding. 

SK Hynix reportedly clarified that the leaked bonding-related information concerned general wafer-to-wafer processes and not the specific hybrid bonding technique currently being developed or used for commercial products. For Chinese companies like Huawei (which does not have access to advanced process technologies and packaging methods used by TSMC and SK hynix) as well as SMIC (which does not have access to the latest lithography equipment made by ASML), getting wafer bonding know-how from SK hynix, which had licensed such technologies from Xperi, may be considered as a major breakthrough as wafer bonding is becoming increasingly important for multiple semiconductor applications.

The individual in question, a South Korean citizen named only in the complaint by his surname, Kim, previously worked for the Chinese unit of SK hynix. Authorities allege that in 2022 he secured a position with HiSilicon and, during the hiring process, improperly accessed and misused internal data belonging to SK Hynix.  Kim reportedly printed out and captured photographs of internal files against policy. Investigators discovered that over 11,000 photographs of confidential content were taken, and steps were allegedly made to mask the origin by removing company identifiers such as brand logos and secrecy labels from the images.  Evidence indicates that Kim submitted these materials as part of job applications to two separate companies in China, using them to enhance his qualifications. Authorities now claim this was a calculated effort to secure employment using unlawfully obtained information from his former employer.  The report confirms that many technologies developed in China in general and by Huawei specifically originate from multi-national corporations. 

Word has reached SemiAccurate about the upcoming ARM branded server CPU. While the picture is not complete yet, we can now flesh out a lot of the details. SemiAccurate has long heard rumors of ARM making their own devices which on the surface appear to compete with some customers. The devices are real, they have customers, and the competition aspect is as unclear as it ever was. All this said, time to dive in.

Sunlin Chou, who passed on December 5th, 2018, played his role and made contributions to the semiconductor industry in developing a new business model for semiconductor R&D. While his legacy had faded since he retired in 2005, his career included filling Gordon Moore’s shoes as a key technologist in the 2nd Gen of Semiconductor Technical Leadership. Sunlin Chou’s last position was running the Technology side of Intel’s Technology and Manufacturing Group. He was in that group of “Shadow Warriors” who ran R&D. They seldom get any public light, and only the people in their company know what they do. Yet they bear a tremendous burden. They must decide among the myriad of technologies, which ones deserve attention, funding, etc. They must ply the foggy seas of the tech triangle that lie between customer need, realization in product, and manufacturability. It is a place where shipwrecks occur all too often. For Sunlin, the challenge was even greater. He had to fill Gordon Moore’s shoes. That meant not only being right, but also being above industry politics. You can learn a lot from someone who has successfully sailed these seas. At the core of Sunlin’s values was the belief that all problems should be dealt with systematically. He was one of the most systematic individuals you would ever meet, exemplary in many ways: every hair was always in a precise place, clothing all in alignment, shoes polished, and a poker stare that came from years of making many calculated gambles on future technology. This is exactly how he dealt with the issue of reviving Intel’s prowess after the 1985 debacle.  Sunlin’s role in this was to develop a new business model for R&D. While many people were looking for an outside solution, like consortia, Sunlin was looking inside to see how the process could be streamlined. Most knew the model wasn’t working well. R&D often never led anywhere for the companies that funded it (Xerox Parc being the best example). It was a money sink, because few innovations ever made it to the business side of an organization. Yet few did anything about it. -- Sunlin looked at these problems and saw the opportunity. -- Sunlin’s most important contribution is his R&D Pipeline model, where he applied the first systematic approach to the whole development cycle. Central to his method was a strong understanding of Moore’s Law and how it impacted product life cycles. It led to his decision to set technical direction based on its relentless progression. This led to node planning. What came next was truly innovative. First, he covered the horizontal time line with a flow model that started with the research infrastructure outside Intel; then feeding it inside to research; then to path finding, on to development; and finally, to manufacturing – all synchronized to the nodal clock. To do this, he had to create a Copy-Exactly infrastructure for R&D that had never existed. Plus, everything had to be tightly coupled at the hand-off points, which is where most fail. This was done with exacting business processes. Second, he also covered the vertical technology line, integrating the test, assembly, and process roadmaps into a single cohesive strategy/roadmap. It doesn’t get more systematic than this. The result was that Intel was able to shift into high gear, running on a 24-month nodal clock.

Data from Canalys shows shipments of desktops and notebooks to the United States grew 15% Y-o-Y in Q1 '25 hitting 16.9M units. A surge in sell-in activity by vendors, driven by efforts to navigate tariff announcements, led to significant inventory buildup that now needs to be cleared. Combined with downward pressure on consumer spending, will lead to a market slowdown for the remainder of 2025, with total PC shipments forecast to increase by ~2%. With a considerable portion of Windows refresh activity set to occur after the Windows 10 End-of-Support deadline in October 2025, growth in 2026 is expected to come in at 4%. “Despite the sense of urgency driven by the Windows 10 EoS date, consumer demand in the US has remained largely unaffected”. “This highlights a key reality: many PC users are indifferent to the Windows 10 EoS until it becomes a significant pain point. Consumer PC purchases tend to be largely driven by factors like poor battery life, slow performance, or hardware failures. Overwhelming specifications and unclear messaging also exacerbate the complexity of the PC buying journey. For vendors and channel partners, this underscores the need to simplify the buyer experience through clear, needs-based messaging and streamlined product portfolios. The channel must proactively guide consumers through a coherent and intuitive purchasing process, helping them easily identify devices tailored to their needs.”

Kevin Zhang, Deputy Co-Chief Operating Officer at TSMC, recently reiterated the company's cautious approach toward adopting High NA EUV lithography tools, emphasizing that TSMC is "in no rush" to bring the advanced technology into high-volume production. His remarks have reignited industry speculation around ASML's multi-billion-dollar next-generation lithography platform.

GlobalFoundries Inc, the biggest US-based provider of made-to-order chips, announced a plan to spend US$16 billion to bolster domestic production.  The company is budgeting US$13 billion to expand existing plants in New York and Vermont and making a further US$3 billion commitment to research into advanced packaging and other technologies in the US, GlobalFoundries is the latest company to publicly commit billions to increasing US production of electronics — announcements. GF said it’s making the investments with the endorsement of customers such as Apple Inc, Qualcomm Inc and General Motors Co. New CEO Tim Breen said GF isn’t providing a detailed breakdown of when the cash will be spent and will remain flexible in order to match supply with demand. The emphasis on the US is “a recognition of where there is the most unmet demand today,” GF says it’s carving out lucrative niches of the semiconductor industry. That includes an effort to combine chips with optical data components and an alternative material for power management chips, called gallium nitride (GaN). GlobalFoundries, is majority-owned by the government of Abu Dhabi, had previously been more conservative in its investment plans, saying it would only add capacity when supported by demand. On average, it’s spent about US$1.4 billion on new plants and equipment annually over the last 5 years. That’s a fraction of the multiple tens of billions that companies like Intel Corp and Samsung Electronics Co have spent.

.

Chip customers are looking for more local production and want to reduce dependence on suppliers that have their manufacturing concentrated in one location, Breen said. “Supply security matters,” he said. Requests for more production out of US plants have increased during the last six months. Globalfoundries, based in Malta, New York, manufactures so-called essential chips for semiconductor and electronics makers.

Such components don’t require the most advanced production, but they do handle vital tasks, such as controlling power and managing the flow of data inside devices. The AI boom also is increasing demand for a variety of chips. The GlobalFoundries investment is “a strategic response to the explosive growth in AI,”. The market has boosted the need for power-efficient and high-bandwidth chips used in Data Centers and communications equipment. 

Globalfoundries was created by merging the former manufacturing operations of AMD with Chartered Semiconductor — a deal that provided the new business with plants in the US, Germany and Singapore.

The company gave up on trying to compete with TSMC in the market for the most advanced production and has instead focused on cheaper techniques needed for other types of chips — components it says are becoming increasingly valuable. 

The report and respondent comments highlight an uncertain operating environment and the negative impacts of trade policies. Supply executives reported a “continued softening of demand” due to “higher prices and economic uncertainty.” The “impact of the current administration’s ever-changing trade policies has wreaked havoc on suppliers’ ability to react and remain profitable.” Some suppliers simply pass tariff costs directly on to customers. Uncertainty surrounding tariffs creates “supply chain disruptions rivaling that of COVID-19,” raising costs, impacting international orders, and forcing companies to undertake extensive and distracting contingency planning without clear direction.  “With the uncertainty, manufacturers need to figure out and forecast profitability based on the cost of materials. (BOM - Bill of Materials Costs) It can be frustrating to have to do it multiple times a month or quarter. Lately, it’s been multiple times a week,” Susan Spence said in a conference call with reporters on Monday. “So, (the current situation) is not benefiting anybody right now, that’s for sure,” adding that navigating policy-driven challenges amid weakening demand will be crucial for the manufacturing sector.

Previous industry sources suggested that NVIDIA was exploring the use of Intel’s 18A process for its gaming GPUs, according to Tom’s Hardware. Now, interest in the 18A node appears to be growing. Chosun Biz notes that Microsoft has reportedly signed a large-scale foundry deal with Intel using the 18A process, while talks with Google are also said to be underway. It's been noted that a Microsoft deal would mark a major milestone for Intel boosting its standing against TSMC and Samsung in advanced foundry. - Intel’s 18A process is currently in risk production and is expected to enter high volume manufacturing (HVM) late 2025. Intel is developing two enhanced variants of its 18A node: 18A-P, slated for launch in 2026, and 18A-PT, expected in 2028, it was noted that initial 18A-P wafers have already been introduced in Intel’s fabs. Intel is working with lead customers on its next-gen 14A node and started sharing early 14A PDKs with key partners.  Chosun Biz notes that Intel has the largest number of U.S.-based factories among foundry players, potentially offering it an advantage in mitigating tariff-related risks.  Intel is currently building 2 cutting-edge fabs in Arizona, alongside an expansion of its advanced packaging facilities in New Mexico. In Oregon, Intel is also moving forward with a new 300mm logic and foundry plant. In Ohio, it has committed USD 28 billion to construct 2 additional factories;  but operations are now expected to begin between 2030 and 2032.

Intel’s overseas expansion is picking up. in Ireland, HVM of the Intel 4 process is set to begin at Fab 34, with 3nm chip HVMN scheduled for later in 2025. In Israel, Fab 38 will produce HPC ICs using EUV lithography. 

Intel is notching up an impressive collection of bugs and flaws of late, crashing 13th and 14th Gen CPUs, the latest "Arrow Lake" chips needing a fair bit of post-launch work to get them running (not all that) optimally, all that good stuff. Now Intel has announced some exciting new bugs that afflict not just its CPUs, but also Arc GPUs.

Let's start with some familiar sounding bugs involving Core Ultra processors. "Description: Incorrect initialization of resource in the branch prediction unit for some Intel Core Ultra Processors may allow an authenticated user to potentially enable information disclosure via local access," via the official security advisory. In other words, the bad guys can gain access to your rig. Intel says it is, "releasing microcode updates to mitigate these potential vulnerabilities," which impact Core Ultra 5, 7, and 9 CPU models across both desktop and laptop. The 2nd CPU bug afflicting those same chips is described as, "a potential security ['flaw'? Even Intel's bug reports have bugs, it seems...] in the Intel Integrated Connectivity I/O interface (CNVi) for some Intel Core Ultra Processors may allow escalation of privilege."  A microcode fix is in the oven. In both cases, Intel advises that PC owners should contact their system provider for an update to fix the problem, but it's not clear if the fix has already been released by Intel.  Next up, are novel GPU flaws, one of which has been given a "HIGH" severity by Intel. "Potential security vulnerabilities in some Intel Graphics Driver software may allow escalation of privilege, denial of service, or information disclosure,".  This applies to all Intel iGPUs from 7th Gen CPUs onwards and also includes the latest Arc GPUs, like the Arc B580. And we have an advisory pertaining to Intel's Endurance Gaming Mode software. "A potential security vulnerability for some Endurance Gaming Mode software may allow escalation of privilege," 

Intel Endurance Gaming Mode is an app for laptops that monitors real-time frame rates. The idea is to keep power consumption in check by setting a target frame rate and then reducing GPU frequency and power to match, resulting in improved battery life. The fix for this is both an update to Endurance Gaming Mode version 1.5.651.0, available here, and that latest GPU driver we mentioned before, available here. All of which means if you have Intel laptop, say, with both the relevant CPU and GPU and you happen to use Endurance Gaming Mode, perhaps you'd better get patched up fast.

Taiwan is investigating whether Semiconductor Manufacturing International Corp. (SMIC), China's largest contract chipmaker and the world's third largest foundry, broke local laws by secretly recruiting engineers to get experienced staff and gain insights of Taiwanese semiconductor know how, reports Bloomberg. The investigation is part of a broader crackdown on unauthorized activities by Chinese tech companies in Taiwan, notes Nikkei. Taiwanese authorities raided 34 different locations in 6 cities between March 18th - 28th. Officials from the Ministry of Justice examined operations linked to 11 Chinese companies and interviewed +90 people in connection with suspected breaches of local laws concerning talent poaching by Chinese companies. Most of the companies under investigation are involved in the semiconductor sector. Officials said this operation had been in the planning stages since December. Around 11 companies, the majority from the semiconductor industry, violated Taiwan's cross-strait regulations. Illegally setting up office without the government's knowledge. Of the 90 interviewees, some were witnesses, while others may have broken the law.  The investigation revealed that SMIC used a company registered in Samoa to discreetly recruit engineers in Taiwan. Authorities also found that several other firms had formed unauthorized local entities to carry out hiring for work on advanced projects, including 'high-performance networking,' which is probably how business media describes silicon photonics, an area that is becoming increasingly important for next-generation AI and HPC datacenters.  SMIC has 7nm-class and preps to roll-out 6nm-class process technologies, the best production nodes available in China. It produces chips for Huawei's HiSilicon, which tends to develop competitive offerings for client devices as well as AI servers and general-purpose servers. SMIC and Huawei need technologies like silicon photonics for their next-generation offerings. However, SMIC is not the only company that allegedly poaches engineers from Taiwanese companies. A company called Clounix was found to have poached a large number of specialists from major companies like Intel and Microsoft. It reportedly first disguised itself as a local company and later as a Singapore-based chip designer to avoid oversight. However, Clounix has financial backing from Ant Group, which is affiliated with Alibaba, which is emerging as a large cloud service provider with its own custom silicon division that develop CPUs and special-purpose ASICs for its broad needs, including general-purpose processing and AI workloads. Interestingly, Chinese entities were not only interested in advanced semiconductor or processor technologies know how. A company that develops display driver ICs has allegedly set up two offices in Tainan and Hsinchu and conducted remote interviews from China to bring in Taiwanese talent. It is unclear whether they wanted Taiwanese engineers to develop DDICs in China.

Currently, Intel's Foundry division loses billions every quarter as it invests heavily in new process technologies and production capacity. However, the company hopes that the Intel Foundry unit will break even sometime in 2027, which will coincide with the rollout out Intel's 14A manufacturing technology and production start on 18A-P node. 
Intel this week reaffirmed that the first product made on its 18A (1.8nm-class) fabrication process, the client PC processor (codenamed Panther Lake), will hit the market late this year and will ramp in 2026. The manufacturing technology will also be adopted for Xeon 'Clearwater Forest' and some 3rd-party products, but from Intel's Foundry business perspective, 18A is will be a proof-of-concept for external clients. If this production node is a success, more potential customers will adopt its successors, including 18A-P, and 14A (1.4nm-class). -- "I think we do need to see more external volume come from 14A versus versus 18A, said David Zinsner, CFO, at the J.P. Morgan Global Technology, Media and Communications Conference.

"We have […] a bunch of bunch of potential customers, and then we get test chips, and then some customers fall out in the test chips, and then there is a certain amount of customers that kind of hang in there. So, committed volume is not significant right now, for sure. But, you know, I think we have got to partly prove ourselves a little bit with our own product and eat our own dog food here, and then […] we start to see some engagement around customers."  Zinsner admitted that if the company choses to use High-NA EUV lithography with its 14A process technology, but its costs will go up initially. Intel hopes that advantages enabled by the new fab tools will outweigh those higher costs.  The company also intends to produce more of its own silicon in house with its Panther Lake and Nova Lake CPUs, which will improve Intel's margins and its utilization of production capacities. As a result, Intel hopes that its Foundry unit will break even in 2027 and will be profitable from then on.  "We still feel on track to to hit breakeven sometime in 2027," Zinsner said. "You know, I think when we committed to it in 2024, we said, 'it is going to be somewhere between 2024 and 2030, most people kind of settled in that that must mean 2027, and that is generally kind of what we are thinking is we can be breakeven."  -- To break even, Intel Foundry only needs low- to mid-single-digit billions in annual external revenue, according to Zinsner. Most of 18A's volume will come from Intel's internal products, while 14A will require more external adoption. Intel's breakeven strategy also includes revenue from advanced packaging, mature nodes (like Intel 16), and partnerships with UMC and Tower. The company will continues to follow its 'smart capital' model, balancing internal and external wafer sourcing, and expects Foundry to compete for internal product demand to ensure efficiency and cost control.

Intel CEO Lip-Bu Tan has laid out the most detailed vision yet for the embattled chipmaker’s turnaround—but offered little clarity to employees on the scope or timeline of anticipated job cuts.
Inside the Intel reckoning: layoffs, leadership, and the long road back to relevance. Tan acknowledged both the company’s modest success in Q1 of 2025 and the significant challenges ahead.  “We are navigating an increasingly volatile and uncertain macroeconomic environment, which is reflected in our Q2 outlook.”  The internal memo followed reports earlier this week from Bloomberg that Intel plans to cut approximately 22,000 jobs, or roughly 20% of its global workforce—one of the largest rounds of layoffs in its history. Tan stated only that “these critical changes will reduce the size of our workforce,” and that the reductions would begin in Q2 2025 and continue “as quickly as possible over the next several months.”
Tan, who took over as CEO five weeks ago, is pushing for deep structural and cultural changes across Intel as it seeks to regain competitiveness amid slowing revenue growth, rising global tariffs, and increasing pressure from rivals in AI and advanced semiconductors.  Tan said in the letter. “We are seen as too slow, too complex and too set in our ways — and we need to change.” A key pillar of Tan’s plan is to refocus the company around engineering, while reducing management layers and streamlining decision-making. “We need to get back to our roots and empower our engineers,”. “That’s why I elevated our core engineering functions to the [Executive Team].”  The letter described Intel’s internal bureaucracy as a drag on innovation. “Many teams are 8 or more layers deep, which creates unnecessary bureaucracy that slows us down,”. He also criticized a corporate culture in which “the most important KPI for many managers … has been the size of their teams.”  Tan announced plans to reduce operating expenses to $17 billion in 2025, down from a previous target of $17.5 billion, and to $16 billion by 2026. CapEx spending will also be cut, with gross capital expenditures lowered from $20 billion to $18 billion in 2025. 
The memo also outlined changes to hybrid work expectations. Starting September 1, employees currently on hybrid schedules will be required to work from the office 4 days/week. The current policy of 3 days/week, has been “uneven at best” in its enforcement. Tan framed the changes as essential to rebuilding Intel’s identity. “Intel was once widely seen as the world’s most innovative company. There’s no reason we can’t get back there, so long as we drive the changes needed to improve,”. However, employees seeking concrete information on how the restructuring will affect them were left with few specifics. Tan emphasized decisions on staffing would be made by his leadership team and shared “on an ongoing basis,” but did not address which divisions or geographies would be affected. “We must balance our reductions with the need to retain and recruit key talent,” he wrote. “These decisions will not be made lightly.”  Intel currently employs around 109,000 people worldwide. The company’s Q1 2025 results showed revenue of $12.67 billion, narrowly exceeding expectations. But its forecast for the next quarter—$11.2 billion to $12.4 billion—fell below analyst estimates and sent shares down more than 7% in after-hours trading.  Intel’s transformation comes amid growing competitive threats from companies like Nvidia and AMD, as well as geopolitical tensions with China, one of its largest markets. “I came on board knowing full well this would be the most challenging job of my career,” he wrote. “But also the most motivating and fulfilling.”