Intel to Acquire Israel-based Granulate

Intel Corporation announced an agreement to acquire Granulate Cloud Solutions, an Israel-based developer of real-time continuous optimization software. The acquisition of Granulate will help cloud and data center customers maximize compute workload performance and reduce infrastructure and cloud costs. Deal terms are not being disclosed, but it is expected to close in the second quarter of 2022. At that time, Granulate’s 120 employees will be integrated into Intel’s Datacenter and AI business unit.

Sandra Rivera, GM of the Datacenter and AI Group at Intel, said that Granulate’s autonomous optimization software can be applied to production workloads without the need to make changes in the customer’s code. Greg Lavender, GM of the Software and Advanced Technology Group at Intel, explained that Granulate’s real-time optimization software complements Intel’s capabilities by helping customers to gain performance and reduce Cloud costs.

How It Works

Granulate’s real-time continuous optimization is a new approach to optimizing production workloads by leveraging resource usage patterns and dataflow to automatically adapt kernel level and runtime level resource management to better fit the application needs. It automatically learns the application’s specific resource usage patterns and data flow to identify contended resources, bottlenecks and prioritization opportunities, and then tailors OS-level scheduling and prioritization decisions to improve the infrastructure’s application specific performance.

Accelerating Legacy Packages

While cloud computing and microservices have created a new era of flexibility in distributed applications and deployment scalability, modern architectures have introduced more complex performance issues that are not easily managed by traditional operating systems and runtimes. Additionally, customers often deploy older Linux distributions and application libraries that are not up to date with the latest advancements in today’s high-performance CPUs.

Granulate’s autonomous optimization service solves these issues by reducing CPU utilization and application latencies, by learning the customer’s application and deploying a customized set of continuous optimizations at runtime. This enables deployment on smaller compute clusters and instance types to improve application performance and drive down cloud and data center costs, without the developer intervention. Thus, optimizations for the latest CPUs can be applied even on legacy Linux distributions and runtimes.

Optimizing Xeon deployments

Asaf Ezra, co-founder and CEO of Granulate, said: “As a part of Intel, Granulate will be able to deliver autonomous optimization capabilities to even more customers globally and rapidly expand its offering with the help of Intel’s 19,000 software engineers.” Intel and Granulate’s relationship began in late 2019, when Granulate was part of the first graduating class of Intel® Ignite, the startup accelerator program that taps into Intel’s resources.

Over the past year, Intel and Granulate have worked together under a commercial agreement to collaborate on workload optimization on Xeon deployments. This collaboration resulted in gains in performance and decreases in costs for customers running on Intel processors. With the acquisition of Granulate, Intel will rapidly scale Granulate’s optimization software, including across Intel’s data center portfolio.

Deci launches new models for enhancing Deep Learning on CPU 

Photo above:Deci’s founders (from left to right): Jonathan Elial- COO, Yonatan Giefman- CEO and Ran El Yaniv- Chief scientist. Credit: Deci

Deci, the deep learning company harnessing Artificial Intelligence (AI) to build AI, announced a new set of image classification models, dubbed DeciNets, for Intel Cascade Lake CPUs. According to Deci, its proprietary Automated Neural Architecture Construction (AutoNAC) technology automatically generated the new image classification models that significantly improve all published models and deliver more than 2x improvement in runtime, coupled with improved accuracy, as compared to the most powerful models publicly available such as EfficientNets, developed by Google.

While GPUs have traditionally been the hardware of choice for running convolutional neural networks (CNNs), CPUs, already more commonly utilized for various computing tasks, would serve as a much cheaper alternative. Although it is possible to run deep learning inference on CPUs, generally they are significantly less powerful than GPUs. Consequently, deep learning models typically perform 3-10X slower on a CPU than on a GPU.

As explained by Deci, its DeciNets closes the gap significantly between GPU and CPU performance for CNNs. With DeciNets, tasks that previously could not be carried out on a CPU because they were too resource intensive are now possible. Additionally, these tasks will see a marked performance improvement: by leveraging DeciNets, the gap between a model’s inference performance on a GPU versus a CPU is cut in half, without sacrificing the model’s accuracy.

“As deep learning practitioners, our goal is not only to find the most accurate models, but  to uncover the most resource-efficient models which work seamlessly in production – this combination of effectiveness and accuracy constitutes the ‘holy grail’ of deep learning,” said Yonatan Geifman, co-founder and CEO of Deci. “AutoNAC creates the best computer vision models to date, and now, the new class of DeciNets can be applied and effectively run AI applications on CPUs.”

All networks were compiled and quantized using OpenVino, with latency measured on AWS instance c5.4xlarge with Cascade Lake CPU (16 vCPUs, batch size = 1)

“There is a commercial, as well as academic desire, to tackle increasingly difficult AI challenges. The result is a rapid increase in the complexity and size of deep neural models that are capable of handling those challenges,” said Prof. Ran El-Yaniv, co-founder and Chief Scientist of Deci and Professor of Computer Science at the Technion – Israel Institute of Technology. The hardware industry is in a race to develop dedicated AI chips that will provide sufficient compute to run such models; however, with model complexity increasing at a staggering pace, we are approaching the limit of what hardware can support using current chip technology. Deci’s AutoNAC creates powerful models automatically, giving users superior accuracy and inference speed even on low-cost devices, including  traditional CPUs.”

In March 2021, Deci and Intel announced a broad strategic collaboration to optimize deep learning inference on Intel Architecture (IA) CPUs. Prior to this, Deci and Intel worked together at MLPerf, where on several popular Intel CPUs, Deci’s AutoNAC technology accelerated the inference speed of the well-known ResNet50 neural network, reducing the submitted models’ latency by a factor of up to 11.8x and increasing throughput by up to 11x.

Deci enables deep learning to live up to its true potential by using AI to build better AI. With the company’s end-to-end deep learning development platform, AI developers can build, optimize, and deploy faster and more accurate models for any environment including cloud, edge, and mobile, allowing them to revolutionize industries with innovative products.  The platform is powered by Deci’s proprietary automated Neural Architecture Construction technology (AutoNAC), which automatically generates and optimizes deep learning models’ architecture and allows teams to accelerate inference performance, enable new use cases on limited hardware, shorten development cycles and reduce computing costs. Founded by Yonatan Geifman, Jonathan Elial, and Professor Ran El-Yaniv, Deci’s team of deep learning engineers and scientists are dedicated to eliminating production-related bottlenecks across the AI lifecycle.

Intel to Acquire Tower for $5.4 Billion

Intel and Tower Semiconductor (based in Migdal Haemek, Israel), announced a definitive agreement under which Intel will acquire Tower for $53 per share in cash, representing a total enterprise value of approximately $5.4 billion. The acquisition supports Intel’s IDM 2.0 strategy to build a foundry services business. “Tower’s specialty technology portfolio, geographic reach and deep customer relationships will help scale Intel’s foundry services and advance our goal of becoming a major provider of foundry capacity globally,” said Pat Gelsinger, Intel CEO.

As a key part of its IDM 2.0 strategy, Intel established Intel Foundry Services (IFS) in March 2021 to become a major provider of U.S.- and Europe-based foundry capacity to serve customers globally. IFS currently offers leading-edge process and packaging technology, and a broad intellectual property (IP) portfolio. The transaction is expected to close in approximately 12 months. It has been unanimously approved by Intel’s and Tower’s boards of directors and is subject to regulatory approvals, including the approval of Tower’s stockholders.

Tower Semiconductor provides a broad range of customizable process platforms such as SiGe, BiCMOS, mixed-signal/CMOS, RF CMOS, CMOS image sensor, non-imaging sensors, integrated power management (BCD and 700V), and MEMS. It owns two manufacturing facilities in Israel (150mm and 200mm), two in the U.S. (200mm), three facilities in Japan (two 200mm and one 300mm) which it owns through its 51% holdings in TPSCo and is sharing a 300mm manufacturing facility being established in Italy with ST.

“With the addition of Tower, Intel is strongly positioned to bring more value to customers across the nearly $100 billion addressable foundry market.”

“Taking Mobileye public is admitting Intel’s failure”

Intel’s intention to take Mobileye public in mid 2022, less than 5 years after its acquisition, continues to resonate and raises questions regarding the direction its current CEO, Pat Gelsinger, is leading the company. In light of the major importance Intel attributed to Mobileye in its future plans, it seems that the expected offering is not a single move; rather it is a sharp u-turn from the strategy led in the past years by the two former CEOs.

In a recent conversation with Techtime, Sergey Vastchenok – Senior Equity Analyst at Oppenheimer Israel explained that the expected offering is the price Intel pays for its mistakes in recent years. “In the recent years, Intel lagged behind its major competitors in the manufacturing world, mainly TSMC and Samsung, with regard to capital expenditures (Capex). This process led Intel to lose its technological advantage”.  

“New CEO, Pat Gelsinger, who grew within Intel and is part of its genetics, came with a vision of reviving its past market dominance, and he did so by massively increasing investments in the company’s organic capabilities”.

Sergey Vastchenok, Oppenheimer Israel

The diminution in Intel’s status is reflected very clearly in its share performance in recent years. In the past 5 years, Intel’s share raised by only 47% whiles the Philadelphia Stock Exchange (PHLX) index, grouping leading chip companies, raised by 354%. Nvidia and AMD, Intel direct competitors, raised by 1300% within the same period.

Vastchenok: “Intel plans to invest 20-25 billion annually in the coming years to restore the lost advantage. Since the company is deficient with cash, the main question is how to finance the planned increase in investments.  Increasing its debt may harm the company’s credit rating, while additional offering in the stock market may weaken its already weak share. The best method of raising funds is to start selling assets”.

Mobileye is a perfect asset to sell

According to Vastchenok, selling Mobileye is a rational step, since its market value is considerably high, and since there was never a real synergy between the two companies’ core activities. Not only there was no real synergy, one may say that it was even a negative one, since Mobileye buy its chips from TSMC and not from Intel. This led to the situation where Intel financed Mobileye’s development strategy, while TSMC benefit the production orders”.   

“Mobileye is a growing business, and is considered a pioneer and a leader in the advanced driver-assistance systems (ADAS) world. Yet, it is still difficult to forecast its revenues and its market share in the future autonomous vehicle world, since there are many competitors nowadays. Every technology giant is joining the game. At this situation, if the market is willing to pay a good price for your assets – you better sell. If Intel is going to make it public at the mentioned price, it could significantly enlarge its cash reserves. This move will do well for both Intel and Mobileye. Mobileye has better chance to prosper outside Intel’s wings. As we can see, since it was acquired by Intel they have lost plenty of quality human resources, the same way Intel lost human resources for its competitors”.

What is the future of Moovit?

Selling Mobileye raises questions regarding the future of other companies bought by Intel in the recent years. In May 2020 Intel acquired Moovit – the Israeli public transport platform – for $900 million. This surprising acquisition presented as another step in Intel’s vision to provide Mobility-as-a-Service (MaaS) services in the future, based on Mobileye’s autonomous vehicles and Moovit’s application. In fact, it seems that the real synergy is between Mobileye and Moovit. “Almost all the acquisitions made by Intel at the past years, to include FPGA vendor Altera, were not related to the core activities of Intel. Taking Mobileye public is an admission of the failure of the acquisition strategy of recent years. It looks like the current CEO is being aware of Intel core expertise and what should they focus at”.

Intel Intends to take Mobileye Public

Above: Autonomous Driving car in front of Mobileye’s HQ in Jerusalem

the Intel announced its intention to take Mobileye public in the United States in mid-2022 via an initial public offering (IPO). Intel will remain the majority owner of Mobileye, and the two companies will continue as strategic partners, collaborating on projects in the automotive sector. The Mobileye executive team will remain, with Prof. Amnon Shashua continuing as the company’s CEO. Recently acquired Moovit as well as Intel teams working on LiDAR, radar development and other Mobileye projects will be aligned as part of Mobileye.

Jerusalem-based Mobileye provides computer vision chips, data analysis, localization and mapping solutions for Advanced Driver Assistance Systems and autonomous driving. On August 2017 Intel completed the acquisition of Mobileye for $15.3 billion. Intel CEO Pat Gelsinger said that the acquisition of Mobileye has been a success. “Mobileye has achieved record revenue year-over-year with 2021 gains expected to be more than 40% higher than 2020. An IPO provides the best opportunity to build on Mobileye’s track record for innovation and unlock value for shareholders.”

In 2021, Mobileye shipped its 100 millionth EyeQ system-on-chip (SoC), scaled autonomous vehicle (AV) test programs across multiple cities around the world covering the U.S., Europe and Asia, unveiled its production robotaxi, and secured 41 new ADAS program wins across more than 30 automakers. Moovit is a Mobility-as-a-Service (MaaS) solutions provider that complements Mobileye’s solutions. The Tel aviv-based company was acquired by Intel on May 2020, for approximately $900 million.

The Mobileye executive team will remain, with Prof. Amnon Shashua continuing as the company’s CEO. Intel currently owns 100% of Mobileye shares and is expected to retain majority ownership following the completion of the IPO. Intel said it has no intention of spinning off or otherwise divesting its majority ownership interest in Mobileye.

Mobileye revenues surged 124% to $ 327 million

Intel’s Mobileye revenue in the second quarter of 2021 totaled $ 327 million, according to the quarterly report released by Intel last week. It represents an increase of 124% year-over-year, but a decrease of 13% compared to the previous quarter. Among all of Intel segments, the Mobileye division posted the highest annual growth in the second quarter .

During the second quarter, Mobileye closed 10 additional design wins in the automotive industry, for over 16 million total lifetime units. The most significant win was with Toyota, in which Mobileye and the german tier-1 supplier ZF  will develop for the Japanese automaker an advanced driver-assistance systems (ADAS) based on a new automotive radar developed by ZF and Mobileye’s computer-vision chip, EyeQ4. The system is intended to be integrated into several of Toyota’s models in the coming years.

Last week, Mobileye began test drives of AVs in New York, the first to receive a permit from the authorities to conduct such trials in the city. In the conference call following the quarterly report, Intel’s CEO Pat Gelsinger said Mobileye is leading the AV industry. “With vehicles in Israel, Germany, Detroit, Tokyo, Shanghai, and coming soon to Paris, Mobileye has the largest global footprint in the AV industry, enabled by our unique REM distributed mapping technology.”

According to Gelsinger by year end there will be over 1 million vehicles providing telemetry for dynamic crowd-source mapping. “It’s a unique and powerful advantage of Mobileye.”

Intel and Mobileye develop LiDAR in a Chip

Mobileye announced during CES 2021 this week that it has produced the first prototype of an Automotive LiDAR sensor based on Intel’s Silicon Photonics technology. The new sensor is planned to be installed in its Autonomouse Driving systems in 2025, when the market will be ready for a mass deployment of AVs. According to Prof. Amnon Shashua, President and CEO of Mobileye, the new sensor is based on Frequency-Modulated Continuous Wave (FMCW) technology and Doppler-style algorithms as opposed to the current Time of Flight sensors.

The FMCW sensor provides 4D velocity relative measurements for a distance of up to 300 meters. Its high resolution detection capability reaches 600 points per degree, created by 2 milliom laser pulses per second (2M PPS). The idea is that current approaches and available sensors are too expensive for consumer AVs. Mobileye needs Radars and LiDARs that are both better and cheaper. To reach L5 autonomy it propose three levels of redundancy in the forward-facing field of view (FoV), and for the rest of the FoV, 2 levels of redundancy.

In this scenario, the EV system consists of 360⁰ camera coverage, 360⁰ Radar cocoon and one forward-facing LiDAR sensor. In fact, Intel owns a unique Fab capable of putting active and passive optical elements on a chip together, including lasers and optical amplifiers, loaded onto a photonic integrated circuit, PIC.

The goals for the future radar chip are also very agressive: It will be a software-defined imaging radar eqipped with 2,304 virtual channels based on 48 by 48 transmitters and receivers. This radar will be able to detect motorcycles beyond 200m, old tire on the road, 140 meters away and low and small hazards on the road.