When it comes to generative AI, the companies that come to mind are Google, Meta, and OpenAI. But dont count Amazon out of the AI race.

Tarek El Moussa's road out of debt to being a millionaire | Your Wallet

One of Amazons biggest focuses is in cloud computing, where the companys Amazon Web Services is a market leader. The online tech and retail giant has also built in-house chips to power the cloud and has partnered with OpenAI rival Anthropic on infrastructure that enables large-scale AI modeling.

Deepak Singh, an AWS vice president, talked to Quartz about how generative AI is affecting Amazon. Here are some highlights from the interview.

Amazon has been using machine learning for a long time, such as the robotics in its fulfillment centers or its cashierless Amazon Go stores, Singh said. There are two areas where Amazon has invested early when it comes to generative AI, Singh said. One is providing the cloud infrastructure for those building AI models. Last year, for example, Amazon launched Bedrock, a service that helps AWS users build and scale AI models. The other area is providing features to help AWS customers build their generative AI applications faster.

In other words, rather than being a leader in whats known as the large-language models that power generative AI products like chatbots, Amazon is focused on providing the infrastructure to house these AI models.

Generative AI very directly was enabled because the cloud exists, Singh said.

Both [Amazon CEO Andy Jassy] and [AWS CEO Adam Selipsky] like to say that we are early in a marathon, he added. We are learning new lessons every day about what large-language models are capable of right now.

If youre a software developer, and especially if youre a new software developer, chances are youre using generative AI as part of your day-to-day experience, Singh said. Large-language models, or LLMs, are good at coding in part because theres a lot of open-source code out there in the world that AI systems can learn from, he said.

Singh added that Amazon employees are starting to use the companys own internal AI chatbots to write marketing copy.

When AWS started, I still had to continue to answer a lot of questions like, why is the bookstore building [cloud] infrastructure? Singh said. You know, I havent heard that one in 15 years now, or 14 years or something like that. Its been a long time. At the same time, companies were more hesitant to migrate to the cloud, which helps businesses run their websites and applications.

But now, Singh said your biggest most stodgy bank has jumped right into the deep end [of generative AI].

But again, he added, I think because theyve adopted the cloud over the last several years, its been easier for them to do that.

Read more:
Amazon executive credits cloud Web Services for AI boom - Quartz

Docker recently announced the general availability of their cloud-based container image builder, Docker Build Cloud. Docker Build Cloud offers a remote shared cache and native builders for AMD64 and ARM64 CPU architectures, with the goal of "improving collaboration" and decreasing image build times.

Docker is a virtualisation technology that provides a loosely isolated environment for an application at runtime. Each environment, called a container, has to be built prior to its execution on a host machine and is distributed as an immutable collection of files called an Image. Each file within the collection is referred to as a layer and is a compressed set of changes to the filesystem. The Docker Builder is responsible for converting a series of instructions in a Dockerfile to an image.

How Dockerfiles get converted to Images(Source: Docker User Guide)

Since Dockers launch in 2013, containers have become a ubiquitous mechanism to achieve the vision of build once, run anywhere, a notion originally made popular by Sun Microsystems to highlight the cross-platform operability of the Java language. A key enabler of this success within the Linux ecosystem has been the use of QEMU, an open-source machine emulator and virtualizer. While this allowed users to build images on a host of one CPU architecture, for example, ARM32v7, and run them on others, it also became a pain point due to the relatively slow emulation process. In late 2019, Dockerintroduced Remote Builders as a mechanism to resolve this challenge.

With Remote Builders, users could register a host machine accessible over mutual transport layer security (mTLS) or secure shell (SSH) as a node for building images. By registering nodes of different target architectures and referencing them during the build stage, users could now build once, or multiple times at once, and run native images without emulation. The major downside of Remote Builders at the time of the functionalitys release was the requirement of its users to set up and maintain a fleet of builder nodes. Docker Build Cloud aims to removethis operational burden by offering them as a service.

How a localDocker client uses a Remote Builder(Source: Docker Remote Builders blog post)

Docker Build Cloud nodes run as Amazon EC2 instances with dedicated EBS volumes. EC2 specification begins with 8vCPUs and 16GB of RAM for customers on Personal and Pro plans, and goes up to 16vCPU with 32GB RAM for Team and Business customers. All Docker plans include a limited number of build minutes shared across an organization, with Docker Build Cloud Team sold as an add-on starting from $5 per user per month. The Docker Build Cloud add-on offers an additional 200 build minutes per user and a 200GB shared cache across the organization. A rough cost estimate of an equivalent remote builder configuration self-hosted on AWS gave a monthly running cost of $850, which suggests that Docker Build Cloud, at $5 per user, is priced favorably towards teams with less than 170 engineers.

AWS Cost Estimate for an equivalent self-hostedconfiguration to Docker Build Cloud (Source: AWS Pricing Estimator)

Finally, Docker Build Cloud is currently only available in the AWS US East region and further information on getting started can be found in the user guide.

Read the rest here:
Docker Build Cloud: Using Cloud Compute and Cache to Decrease Image Build Times - InfoQ.com

Rome, 19/02/2024 Supercomputers, artificial intelligence, and cloud are aboard a constellation of cyber-secure satellites orbiting the Earth this is the objective of the Military Space Cloud Architecture (MILSCA) study project assigned to Leonardo by the Italian Ministry of Defense (through the contractual agency Teledife), as part of the National Military Research Plan (PNRM).

For the first time in Europe, similar to what happens with the terrestrial cloud, the project intends to define a space architecture capable of providing government and national Armed Forces with high-performance computing and storage capacity directly in space.

The system, designed with integrated cyber security models, will guarantee greater speed and flexibility in the processing and sharing information. The Space Cloud, which will be tested by creating a digital twin of the architecture, will be able to store over 100 Terabytes of data generated on Earth and in space on board each constellation satellite. It can perform processing with a power exceeding 250 TFLOPS (250 thousand billion operations per second) at single precision, adopting advanced algorithms which use artificial intelligence, machine learning techniques, and extensive data analysis. They can also communicate and exchange data autonomously with other satellites.

A cyber-secure supercomputer and archive system in space will guarantee users access to strategic data such as communication, earth observation, and navigation data, anywhere, even in the most remote places, and at any time. Furthermore, a Space Cloud system significantly reduces data processing times, which is processed directly in orbit, providing real-time information, and thus facilitating multi-domain and multi-nation operations. The transmission networks will be left free for other connections thanks to the only transfer of information of interest to Earth. In addition, storing data in orbit will also represent a useful back-up of the Earth centers, which are most exposed to natural disasters.

The project sees Leonardo at the forefront with the participation of the joint ventures Telespazio (67% Leonardo, 33% Thales) and Thales Alenia Space (67% Thales, 33% Leonardo). With a duration of 24 months, the study includes a first phase for defining the architecture and a second phase that will end with developing a digital twin of the satellite with the HPC and the multi-constellation satellite terminal demonstrator. The goal is to simulate the different application scenarios in a digital environment. These tests will be carried out thanks to Leonardos supercomputer, the davinci-1, among the first aerospace and defense HPCs in the world in terms of computing power and performance. The study will be a precursor to a further experimental phase, which, if confirmed, will involve the deployment in orbit of a demonstrative constellation of satellites.

Space Cloud is a hi-tech and multi-domain project, which takes advantage of Leonardos combined capabilities in data acquisition, management, and cyber protection, as well as artificial intelligence and supercomputing with the HPC davinci-1; the development of MILSCA is the first project in the Space domain that fits within the growth guidelines of Leonardos new Industrial Plan.

In a multi-domain scenario, management, security, and rapid exchange of an ever-increasing amount of data, much of which is tactical, become strategic elements for the countrys defense. We will be the first in Europe to develop a Space Cloud project, demonstrating feasibility and benefits deriving from the use of an architecture of this type and enabling a new paradigm of cloud & edge computing, said Simone Ungaro, Leonardos Chief Innovation Officer. Leonardos know-how will allow the development of a Space Cloud network to contribute to digitalization and technological innovation processes, responding to future challenges to guarantee the needs of government and national Armed Forces.

The Space Cloud for Defense project also sets the basis for future uses to support civil Earth observation programs and space exploration missions to the Moon and Mars, which could benefit from an in-orbit cloud computing architecture to download and process data more quickly.

Image courtesy Leonardo

See the original post here:
Leonardo: kick off for the project of the first Space Cloud System for Defense - EDR Magazine

With its highly successful A100 and H100 processors for artificial intelligence (AI) and high-performance computing (HPC) applications, NVIDIA dominates AI datacenter deployments these days. But among large cloud service providers as well as emerging devices like software defined vehicles (SDVs) there is a global trend towards custom silicon. And, according to a report from Reuters, NVIDIA is putting together a new business unit to take on the custom chip market.

The new business unit will reportedly be led by vice president Dina McKinney, who has a wealth of experience from working at AMD, Marvell, and Qualcomm. The new division aims to address a wide range of sectors including automotive, gaming consoles, data centers, telecom, and others that could benefit from tailored silicon solutions. Although NVIDIA has not officially acknowledged the creation of this division, McKinneys LinkedIn profile as VP of Silicon Engineering reveals her involvement in developing silicon for 'cloud, 5G, gaming, and automotive,' hinting at the broad scope of her alleged business division.

Nine unofficial sources across the industry confirmed to Reuters the existence of the division, but NVIDIA has remained tight-lipped, only discussing its 2022 announcement regarding implementation of its networking technologies into third-party solutions. According to Reuters, NVIDIA has initiated discussions with leading tech companies, including Amazon, Meta, Microsoft, Google, and OpenAI, to investigate the potential for developing custom chips. This hints that NVIDIA intends to extend its offerings beyond the conventional off-the-shelf datacenter and gaming products, embracing the growing trend towards customized silicon solutions.

While using NVIDIA's A100 and H100 processors for AI and high-performance computing (HPC) instances, major cloud service providers (CSPs) like Amazon Web Services, Google, and Microsoft are also advancing their custom processors to meet specific AI and general computing needs. This strategy enables them to cut costs as well as tailor capabilities and power consumption of their hardware to their particular needs. As a result, while NVIDIA's AI and HPC GPUs remain indispensable for many applications, an increasing portion of workloads now run on custom-designed silicon, which means lost business opportunities for NVIDIA. This shift towards bespoke silicon solutions is widespread and the market is expanding quickly. Essentially, instead of fighting custom silicon trend, NVIDIA wants to join it.

Meanwhile, analysts are painting the possibility of an even bigger picture. Well-known GPU industry observer Jon Peddie Research notes that they believe that NVIDIA may be interested in addressing not only CSPs with datacenter offerings, but also consumer market due to huge volumes.

"NVIDIA made their loyal fan base in the consumer market which enabled them to establish the brand and develop ever more powerful processors that could then be used as compute accelerators," said JPR's president Jon Peddie. "But the company has made its fortune in the deep-pocked datacenter market where mission-critical projects see the cost of silicon as trivial to the overall objective. The consumer side gives NVIDIA the economy of scale so they can apply enormous resources to developing chips and the software infrastructure around those chips. It is not just CUDA, but a vast library of software tools and libraries."

Back in mid-2010s NVIDIA tried to address smartphones and tablets with its Tegra SoCs, but without much success. However, the company managed to secure a spot in supplying the application processor for the highly-successful Nintendo Switch console, and certainly would like expand this business. The consumer business allows NVIDIA to design a chip and then sell it to one client for many years without changing its design, amortizing the high costs of development over many millions of chips.

"NVIDIA is of course interested in expanding its footprint in consoles right now they are supplying the biggest selling console supplier, and are calling on Microsoft and Sony every week to try and get back in," Peddie said. "NVIDIA was in the first Xbox, and in PlayStation 3. But AMD has a cost-performance advantage with their APUs, which NVIDIA hopes to match with Grace. And since Windows runs on Arm, NVIDIA has a shot at Microsoft. Sony's custom OS would not be much of a challenge for NVIDIA."

The rest is here:
Report: NVIDIA Forms Custom Chip Unit for Cloud Computing and More - AnandTech