1. The U.S. Department of Commerce cancels $7.4 billion semiconductor research funding, accusing the Biden administration of corruption.2. Skyworks appoints former AMD executive Philip Carter as CFO, with a base salary of $600,000.3. Saudi AI startup Humain will operate its first data centers next year, using NVIDIA chips.4. TSMC’s U.S. packaging plant begins site preparation, aiming for operation in 2028.5. Malaysia launches its first self-developed AI chip MARS1000, joining the global tech competition.6. Connect, perceive, reason, repeat: How Ceva drives the entire intelligent edge stack.
1. The U.S. Department of Commerce cancels $7.4 billion semiconductor research funding, accusing the Biden administration of corruption.
On Monday (August 25), the U.S. Department of Commerce announced that it would cancel a previously approved semiconductor research grant agreement worth up to $7.4 billion, which was to be managed by the National Semiconductor Technology Center (NSTC) under the National Institute of Standards and Technology (NIST).
The Department of Commerce stated that the NSTC operator Natcast (National Semiconductor Technology Promotion Center), established by the Biden administration, violated legal norms and effectively became a “semiconductor slush fund” controlled by several former Biden officials. U.S. Secretary of Commerce Howard Lutnick stated that since its establishment, Natcast has “only allowed the Biden camp to enrich itself with taxpayer funds.”
The Department of Commerce indicated that the Biden administration illegally created Natcast, thus rendering the agreement granting the organization up to $7.4 billion in taxpayer funds invalid.
The Department of Commerce stated that the operation of the semiconductor technology center will undergo reforms and comply with the requirements of the $52.7 billion CHIPS Act.
2. Skyworks appoints former AMD executive Philip Carter as CFO, with a base salary of $600,000.
Skyworks Solutions announced that Philip Carter will return to the chip manufacturer as Chief Financial Officer (CFO) starting September 8.
Philip Carter will leave AMD, where he served as Vice President and Chief Accounting Officer since November 2024.
Skyworks stated that Philip Carter has notified AMD of his resignation and added that his departure is not due to any disagreements regarding accounting practices or financial reporting.
Philip Carter previously worked at Skyworks from 2017 to 2024, serving as Vice President and Corporate Financial Director.
Philip Carter will succeed Robert Schriesheim, a board member who has been serving as interim CFO since May of this year.
Robert Schriesheim will remain on the board.
At Skyworks, Philip Carter’s base salary is $600,000, plus an annual cash bonus targeted at 100% of his base salary.
Skyworks will pay an $800,000 signing bonus, which is to be paid after Philip Carter has been employed for 24 months.
Skyworks is a leading manufacturer of high-performance analog semiconductors with significant influence in the wireless communications field. It enables seamless connectivity for smartphones, automobiles, IoT devices, and more through the design and production of critical RF and mixed-signal semiconductor devices.
On August 5, Skyworks announced its financial results for the quarter ending June 27, 2025.
According to the report, Skyworks’ third-quarter revenue was $965 million, with an operating profit of $111 million, gross profit of $401 million, net profit of $105 million, and diluted earnings per share of $0.70; year-to-date operating cash flow was $1.1 billion, with free cash flow of $962 million.
3. Saudi AI startup Humain will operate its first data centers next year, using NVIDIA chips.
The newly established AI company Humain in Saudi Arabia has begun constructing its first data centers in the country and plans to operate them using semiconductors imported from the U.S. by early 2026.
Humain CEO Tareq Amin stated that the data centers located in Riyadh, the capital of Saudi Arabia, and the eastern province of Dammam are expected to be operational by the second quarter, with an initial capacity of up to 100 megawatts for each data center.
Tareq Amin mentioned that Humain is currently procuring the necessary semiconductors for its data centers from U.S. chip manufacturers, including NVIDIA’s latest AI chips, which have been approved by local regulatory authorities.
In May, NVIDIA announced it would sell hundreds of thousands of AI chips in Saudi Arabia, with the first 18,000 units of the latest “Blackwell” chips being sold to Humain.
In May, several U.S. tech companies announced AI deals in the Middle East, and U.S. President Donald Trump secured a $600 billion investment commitment from Saudi Arabia to U.S. companies during his visit to the Gulf states.
AMD also announced an agreement with Humain, stating that it has established a $10 billion partnership.
Humain was established in May under the Saudi Public Investment Fund (PIF) and is chaired by Saudi Crown Prince Mohammed bin Salman. The company provides AI services and products, including data centers, AI infrastructure, cloud capabilities, and advanced AI models.
4. TSMC’s U.S. packaging plant begins site preparation, aiming for operation in 2028.
TSMC’s expansion in the U.S. is accelerating, with the latest industry news indicating that the two advanced packaging plants (AP1, AP2) planned by TSMC in the U.S. have entered the site preparation phase, with construction expected to begin in the second half of 2026 and aiming for official operation in 2028.
It is reported that in TSMC’s process planning in the U.S., AP1 will introduce the most advanced SoIC and CoW technologies, while AP2 will focus on CoPoS technology to meet local packaging needs for AI and high-performance computing (HPC) chips.
Among them, SoIC is currently the most advanced packaging technology that TSMC has mass-produced, and it will be integrated with the later-stage CoWoS and even future CoPoS. Currently, customers include AMD, which has achieved mass production, as well as Apple, NVIDIA, and Broadcom, all of which will adopt it in high-end products.
Previously, TSMC stated that the third wafer fab (P3) in the U.S. will adopt N2 and A16 process technologies, while the fourth wafer fab (P4) will also use the same technologies, and the fifth and sixth wafer fabs (P5, P6) will adopt even more advanced technologies. The construction and mass production plans for these fabs will depend on customer demand.
Industry insiders pointed out that the construction speed of packaging plants is usually faster than that of wafer fabs. The U.S. packaging plant has begun site preparation, and construction is expected to officially start in the second half of next year, with related equipment expected to be installed in the first half of 2028, and small-scale production anticipated by the end of that year.
5. Malaysia launches its first self-developed AI chip MARS1000, joining the global tech competition.
Malaysia has released its first self-developed artificial intelligence (AI) processor, joining the global competition to create the most sought-after AI electronic components.
The local chip design company SkyeChip launched the MARS1000 chip at an industry association event attended by senior Malaysian government officials. The Malaysian Semiconductor Industry Association stated in a statement that this chip is Malaysia’s first edge AI processor, meaning that it can support devices ranging from automobiles to robots from within.
This Southeast Asian country is seeking to play a larger role in the global chip supply chain and leverage the AI boom. Malaysia is already a significant player in the global semiconductor packaging sector and has become a manufacturing hub for semiconductor equipment suppliers, including Lam Research. Additionally, Malaysia is a thriving center for AI data centers, attracting significant investments from companies like Oracle and Microsoft.
The complexity and performance of edge AI chips are far inferior to NVIDIA’s cutting-edge products that support data centers and train algorithms on a large scale, but it is still a crucial step in building advanced technological capabilities. It is currently unclear where SkyeChip will produce its designed chips.
Malaysia aims to enhance its strengths in chip design, wafer manufacturing, and AI data centers. The government, led by Prime Minister Anwar Ibrahim, has committed to investing at least 25 billion ringgit (approximately $6 billion) to improve its position in the global value chain.
The Trump administration proposed restrictions on the flow of AI chips to Malaysia and Thailand, complicating this effort, as the U.S. suspects that smugglers are using these two countries as transit points to transfer semiconductors to restricted markets. Malaysia has recently taken action to tighten AI chip exports in cooperation with U.S. tech companies and has stated that it will not tolerate the abuse of its country for illegal trade activities.
6. Connect, perceive, reason, repeat: How Ceva drives the entire intelligent edge stack.
The intelligent edge is no longer just a vision for the future; it has arrived today and has become the functional foundation of our daily mobility. From connected cars and smart homes to headphones, wearables, and industrial systems, intelligence is gradually moving closer to where data is generated and decisions are made.
In recent weeks, we have explored how Ceva helps customers build intelligence at the edge through trusted IP—achieving wireless connectivity, contextual awareness, and AI reasoning. As the intelligent edge accelerates its evolution, Ceva’s executive team has proposed a clear blueprint—intelligence, connectivity, and perception converge directly in devices, freeing them from reliance on the cloud, thus enabling immediate responses and immersive contextual experiences.
At the core of Ceva’s strategy is the injection of powerful capabilities into the intelligent edge through a comprehensive and unique portfolio of licensable IP. This goes far beyond simply cramming intelligence into devices. The key is to empower customers to launch differentiated, efficient, and future-oriented products, ensuring they are scaled and delivered on time to market.
From cloud dependency to edge autonomy
In our first article, we pointed out a core shift the industry is experiencing. As more and more data is generated at the edge—from wearables to vehicles—the “cloud-first” model has shown its limitations. Latency, privacy, energy efficiency, and cost pressures all demand a new solution: executing AI reasoning at the edge.
Against the backdrop of the intelligent edge market expected to exceed $127 billion, with reasoning demand surpassing that of the cloud, the question is not “why do edge AI,” but “how to achieve it.” Ceva’s answer is embedded intelligence—real-time decision-making supported by domain-customized IP and scalable architectures, rather than relying solely on general-purpose cores.
Connectivity: The silent driver of the intelligent edge
The capabilities of edge devices depend on their communication capabilities. This is precisely the theme we highlighted in our second article: the core role of multi-protocol wireless connectivity in mission-critical scenarios. From Bluetooth 6 and Wi-Fi 7 to UWB and cellular IoT, seamless interconnectivity across protocols is indispensable.
However, connectivity remains a complex and non-differentiated feature that is difficult to develop in-house. Therefore, Ceva’s wireless IP—adopted by 10 of the top 13 MCU manufacturers globally and shipping over 1 billion devices annually—has become the de facto standard for companies needing certified, scalable, and energy-efficient solutions.
Perception: The cognition that changes experiences
Edge intelligence requires data, and the starting point for data is the perception of the surrounding environment. In our third article, we explored how multimodal perception enables devices to have true responsiveness—capable of understanding motion, sound, location, and context.
Whether it’s headphones automatically adjusting sound effects based on head movements or automotive systems recognizing drivers and personalizing cabin experiences, Ceva’s integrated perception technologies—MotionEngine™, RealSpace®, ClearVox™, and AI-enhanced DSP—bring this vision to life. The combination of perception, connectivity, and AI is key to elevating devices from “smart” to “intuitive.”
At the edge: Focusing on reasoning, not just computational architecture
In our recent fourth article, we returned to the core topic of edge AI—reasoning. Today’s neural network models, whether for vision or language, require not only powerful computing power but also architectures designed for efficiency, flexibility, and energy savings.
Ceva’s NeuPro series perfectly meets this demand—ranging from the ultra-light NeuPro Nano for TinyML applications to the high-performance NeuPro-M core suitable for ADAS and smart cameras. Our unique big-little core architecture achieves the best balance between performance and power consumption, combined with development kits and a unified SDK to accelerate deployment and optimization, compatible with mainstream frameworks and supporting Ceva’s proprietary model library.
Strategic synergy: A platform, not a point solution
In a highly fragmented market, Ceva’s strategy resonates due to its overall synergy. As highlighted by ABI Research’s latest analysis of edge AI, the future is no longer about isolated specifications but about platforms capable of achieving real deployments across different form factors, workloads, and verticals.
This is where Ceva’s advantage lies. Our unified IP platform enables customers to integrate connectivity, perception, and reasoning capabilities into various applications, from industrial sensors and automotive systems to wearable audio devices and smart homes. By 2024, the number of devices driven by Ceva is expected to exceed 2 billion, with total shipments surpassing 19 billion. We are not only driving the realization of the intelligent edge but also continuously expanding its scale.
Looking to the future: From fragmentation to unified foundations
The rapid development of edge AI is reshaping how devices connect, perceive, and respond, but it is also making systems increasingly complex. Talent shortages, lack of standardization, and rising system pressures could hinder progress. At Ceva, we believe the answer lies in a foundational integrated approach: trusted IP, solid ecosystem partnerships, and a simplified path focused on scalable deployment.
For over thirty years, Ceva has continuously empowered industry innovation with core technology modules in connectivity, perception, and intelligence. Our solutions help customers simplify complexity—accelerating development while reducing risk and unlocking differentiated value.
This approach is now being realized, driving edge intelligence:
● In the automotive sector, our embedded AI and sensor fusion platform creates a more personalized, context-aware in-car experience—including low-power voice assistants and adaptive driving preferences.
● In the headphone sector, Ceva’s Bluetooth and AI DSP technologies empower over half of the non-Apple headphones globally, achieving adaptive noise cancellation, voice commands, and gesture recognition, with excellent energy efficiency.
● In smart home scenarios, our connectivity IP seamlessly bridges protocols like Wi-Fi, Bluetooth, and Zigbee, while integrating audio and motion sensing to make homes smarter and more intuitive.
These are just some examples of how we help customers build smarter, faster, and more powerful edge devices.
As the boundaries between cloud and edge dissolve, systems require not just connectivity but also the ability to understand, judge, and execute locally. Ceva’s mission is to drive this transition. We play the role of the behind-the-scenes “engine,” helping customers build the next generation of edge intelligence quickly, at scale, and with confidence.
The edge is no longer a fragmented challenge—it is the cornerstone of competitive advantage. With Ceva, this cornerstone is already in place. (CEVA IP)
Hot Topics:
1.President of domestic chip leader under investigation, what happened?2.Breaking! Two domestic semiconductor companies go bankrupt!3.Chinese smartphone giant banned from sales in India!4.Is the era of “chip for equity” beginning? Details of U.S. government investment in Intel revealed.5.CEO absconds! AI company valued at $7 billion seeks acquisition.
6.Domestic lithography equipment, batch introduction to leading enterprises!
7.Reports suggest NVIDIA is building stronger AI chips for China, outperforming H20.
8.Suspended! Huahong Semiconductor announces acquisition of Huali Micro;
Share
Like
Watching