Wednesday, June 19, 2024

Astera demos PCIe optical connectivity for expansive GPU clusters

Astera Labs has demonstrated end-to-end PCIe optical connectivity to provide unprecedented reach for larger, disaggregated GPU clusters.

The company’s technology for PCIe over optics expands its Aries family of Smart DSP Retimers and Smart Cable Modules (SCMs) for Active Electrical Cables (AECs) to deliver PCIe and CXL connectivity in chip-to-chip, box-to-box, rack-to-rack, and now row-to-row topologies throughout the data center.

Astera Labs said its new technology for PCIe over optics operates with end-to-end PCIe link connectivity in compliance with current PCIe specifications and is flexible enough to address future standards requirements through software-enabled configurability. The company is collaborating with leading optical cable suppliers to enable a wide variety of industry-standard form factors.

In addition, Astera Labs’ COSMOS software suite has been enhanced to provide link, fleet, and RAS management features over optical links, delivering deep insights into every lane and link inside a cluster to facilitate maximum uptime and optimal infrastructure utilization. Customers already using Astera Labs’ Intelligent Connectivity Platform will be able to seamlessly and transparently enable optical interconnects.

Casey Morrison, Chief Product Officer at Astera Labs, said, “GPU clusters are growing in size to handle the expanding complexities of AI workloads within disaggregated cloud infrastructure. We’re proud to broaden our PCIe leadership once again by demonstrating robust PCIe links over optical interconnects between GPUs, CPUs, CXL memory devices, and other PCIe endpoints. This breakthrough expands our Intelligent Connectivity Platform to allow customers to seamlessly scale and extend high bandwidth, low latency PCIe interconnects over optics, including Active Optical Cables, in addition to copper-based AECs.”

Large AI models benefit from tightly clustering more GPUs over a backend mesh network, often using PCIe interfaces that are natively found on GPUs. Larger GPU clusters must be distributed beyond a single rack to meet power delivery and cooling requirements. PCIe optical interconnects are essential to scale PCIe-based GPU clusters to multiple racks and rows, improve AI model performance, and increase GPU utilization.





undefined

Black Semiconductor raises EUR 254.4 million for graphene PICs

Black Semiconductor, a start-up based in Aachen, Germany, announced €254.4 million in funding, marking one of the largest investments for a deep tech company manufacturing chips in Europe to date.

The company is targeting optical chip-to-chip connections using graphene as an intermediate layer to integrate optics with electronics. Black Semiconductor asserts that graphene, a material just one atom thick and known for its superior optical, electrical, and thermal properties, is key to enabling the co-integration of electronics and optics. "Graphene outperforms any other material system, opening the door to products that were previously inaccessible," the company stated. "Our solution combines computing with electrons and communication with photons, which we believe could enable parallel optical chip connectivity over distances of up to kilometers."

Black Semiconductor has secured €228.7 million in public funding from the German Ministry of Economic Affairs and Climate Action and the state of North Rhine-Westphalia over the next seven years under the IPCEI ME/CT2 program. Additionally, the company has raised €25.7 million in equity funding. This round was led by Porsche Ventures and Project A Ventures, with participation from leading venture capital firms, corporates, and industry leaders including Scania Growth Capital, Capnamic, Tech Vision Fonds, and NRW.BANK, joining the seed round investors Vsquared Ventures, Cambium Capital, and Hermann Hauser’s Onsight Ventures.

Founded in 2020 by Dr. Daniel Schall and Sebastian Schall, Black Semiconductor plans to use the new capital to accelerate its R&D initiatives and establish pilot line manufacturing capabilities in Aachen by 2026—an essential step towards its long-term goal of producing and implementing high-quality graphene in Europe. The company also aims to increase its headcount from 30 to 120 by 2026.

https://blacksemi.com/

undefined

Global Telco AI Alliance builds Multilingual Telco LLM

SK Telecom, Deutsche Telekom, e&, Singtel, and SoftBank Corp., the founding members of the Global Telco AI Alliance (GTAA), have signed a Joint Venture agreement at TM Forum's DTW24-Ignite. This follows the announcement made at MWC Barcelona 2024 to establish a Joint Venture, marking the founding parties' commitment to co-develop and launch multilingual Large Language Models (Telco LLM) specifically tailored for telecommunications companies. The Telco LLM aims to enhance customer interactions through digital assistants and other innovative AI solutions.

The Joint Venture will see equal investments from the founding parties to support its initial working capital requirements. The Telco LLM will cater to a global customer base of approximately 1.3 billion people across 50 countries, offering multilingual capabilities in languages such as Korean, English, German, Arabic, and Bahasa, among others. The launch of the Joint Venture is pending customary regulatory approvals. The focus will be on deploying innovative AI applications tailored to the unique needs of each founding party's market.

The second Global Telco AI Roundtable (GTAR) was also held at DTW24, emphasizing the GTAA’s dedication to innovation and collaboration in the telecom industry. SKT’s CEO Ryu Young-sang highlighted the transformative potential of AI in telecoms, positioning the Global Telco AI Alliance as a key player in the global AI ecosystem. He stressed the importance of creating new business opportunities and enhancing customer experiences through the Telco LLM JV while addressing social and environmental responsibilities through effective AI governance frameworks.

Key Points:

  • SK Telecom, Deutsche Telekom, e&, Singtel, and SoftBank Corp. sign Joint Venture agreement at TM Forum's DTW24-Ignite.
  • Joint Venture to develop multilingual Large Language Models (Telco LLM) tailored for telecommunications.
  • Equal investments from founding parties to support initial working capital requirements.
  • Telco LLM to cater to a global customer base of 1.3 billion across 50 countries.
  • Second Global Telco AI Roundtable reaffirms GTAA’s commitment to innovation, collaboration, and effective AI governance.

undefined

SNS Telecom & IT: Private 5G market to grow at 42% CAGR to 2027

 Private LTE networks have been a niche market for over a decade, with early installations like iNET's 700 MHz LTE network in the Permian Basin and Rio Tinto's network in Western Australia. However, private cellular networks based on the 3GPP-defined 5G standard are now advancing from Proof-of-Concept (PoC) trials to production-grade implementations. These standalone 5G networks are set to lay the foundation for Industry 4.0 and advanced application scenarios, offering significant improvements over LTE in terms of throughput, latency, reliability, and connection density. Despite initial challenges, early adopters are seeing tangible benefits, affirming the long-term potential of private 5G networks, according to a recently published study by SNS Telecom & IT.

SNS Telecom & IT estimates that annual investments in private 5G networks for vertical industries will grow at a CAGR of approximately 42% between 2024 and 2027, eventually accounting for nearly $3.5 Billion by the end of 2027. 

Compared to LTE, private 5G networks can meet more demanding performance requirements. They offer capabilities like Ultra-Reliable, Low-Latency Communications (URLLC) and Massive Machine-Type Communications (mMTC), making them a viable alternative to wired connections for industrial-grade communications. Despite higher costs, 5G's wider coverage, scalability, and security features have attracted strong interest for use in Industrial IoT (IIoT) environments. China leads in this adoption, with large-scale private 5G installations supporting complex industrial operations. As digitization initiatives ramp up globally, private 5G networks are being implemented for diverse use cases, showing practical benefits in efficiency, cost savings, and worker safety.

Some Highlights

  • Private LTE networks have been in use for over a decade, particularly in niche segments like mining and offshore infrastructure.
  • Private 5G networks are moving beyond PoC trials to full-scale deployments, providing the foundation for Industry 4.0.
  • Private 5G networks offer superior performance in terms of throughput, latency, reliability, and connection density compared to LTE.
  • 5G's URLLC and mMTC capabilities support industrial-grade communications, making it a viable alternative to wired connections.
  • China leads with large-scale private 5G installations, while other countries are also increasing their adoption for digitization and automation.
  • Diverse Use Cases: Applications include wirelessly connected machinery, autonomous mobile robots, augmented reality, remote-controlled cranes, and digital twin models.
  • Challenges: Early adopters face challenges like high costs, limited device variety, and lack of cellular wireless systems competence.
  • Long-Term Potential: Despite initial challenges, early adopters are seeing tangible benefits, affirming the long-term potential of private 5G networks.

The “Private 5G Networks: 2024 – 2030 – Opportunities, Challenges, Strategies & Forecasts” report presents an in-depth assessment of the private 5G network market, including the value chain, market drivers, barriers to uptake, enabling technologies, operational and business models, vertical industries, application scenarios, key trends, future roadmap, standardization, spectrum availability and allocation, regulatory landscape, case studies, ecosystem player profiles and strategies. The report also presents global and regional market size forecasts from 2024 to 2030. The forecasts cover three infrastructure submarkets, 16 vertical industries and five regional markets.

More info here: https://www.snstelecom.com/private5g


undefined

Datagrid surveys route for subsea cable linking NZ and Australia

Datagrid New Zealand has initiated a research project with international marine survey company EGS Survey to find the shortest and most secure route for the proposed 2,700km Te Waipounamu submarine cable between the South Island of New Zealand and Australia, linking Invercargill to Sydney and Melbourne. Datagrid, which is developing a 43-hectare sustainable data center park in Makarewa, aims to provide direct connectivity to Australia, enhancing internet speeds and digital infrastructure for the South Island.

The Te Waipounamu submarine cable is expected to significantly boost the digital development of cities like Christchurch, Dunedin, Queenstown, Nelson, and Invercargill, transforming Invercargill into a regional digital hub. The cable will also enhance New Zealand's network resiliency by offering backup to existing North Island cables. Perrine Dhalluin, CEO of Datagrid, emphasized the strategic importance of the cable for faster internet and regional digital growth. EGS Survey, with 48 years of geophysical surveying experience, will conduct a detailed desktop study to determine the optimal route across the Tasman Sea, ensuring maximum diversity and reliability. An optional branch to Hobart, Tasmania, is also being considered to provide further connectivity options.

Key Points:

  • Companies Involved: Datagrid New Zealand and EGS Survey.
  • Project Goal: To find the shortest and most secure route for the Te Waipounamu submarine cable.
  • Significance: First international submarine cable connecting South Island, NZ, to Australia.
  • Impact: Enhanced internet speed, digital development, and network resiliency for the South Island.
  • Additional Consideration: Potential branch to Hobart, Tasmania, for increased connectivity.
  • The project is named after the Maori term for the South Island


undefined

TM and Singtel's Nxera plan 200 MW data center campus near Singapore

 TM (Telekom Malaysia) has announced a strategic joint venture with Nxera, the regional data center arm of Singtel’s Digital InfraCo unit, to develop state-of-the-art data centers in Malaysia. The first project under this partnership is a sustainable, hyper-connected AI-ready data center campus in Johor. This venture aims to cater to the growing demands of hyperscalers, next-generation AI application providers, and enterprises pursuing accelerated digitalization and cloud services in the region. Leveraging TM's extensive subsea cable networks and Singtel’s global connectivity, the collaboration promises enhanced network performance with lower latency and improved reliability.

The data center campus, located in Iskandar Puteri, just 16 km from Singapore, will be the largest for both TM and Nxera. It is designed to meet the highest standards of reliability, security, and sustainability, offering superior local and international connectivity. The proximity to subsea cable links between Singapore and Malaysia positions it well to support the increasing digitalization efforts of both nations. The initial phase of the data center will have a capacity of 64MW, scalable up to 200MW, featuring advanced technologies like liquid cooling to support high-power density workloads efficiently. The facility will also be LEED-certified, underscoring its commitment to energy efficiency and sustainable practices.

Highlights

  • Strategic Partnership: TM partners with Nxera to develop hyper-connected AI-ready data centers in Malaysia, starting with a major facility in Johor.
  • Advanced Infrastructure: Initial phase of 64MW, scalable to 200MW, featuring liquid cooling and LEED certification for energy efficiency.
  • Enhanced Connectivity: Proximity to subsea cable links ensures superior local and international connectivity.
  • TM’s Data Centers: TM operates seven data centers across Malaysia, with key locations in Klang Valley and Johor.
  • Nxera’s Expansion: Nxera, backed by Singtel and KKR, is developing additional AI-ready data centers, increasing its capacity to over 200MW in the next three years.
  • This partnership marks a significant step forward in enhancing Malaysia’s data infrastructure, positioning the country as a key player in the regional digital economy.

undefined

Three UK gains energy efficiency at base stations with Ericsson

Three UK has reached a significant milestone in network sustainability through the deployment of next-generation AI-powered hardware and software solutions from Ericsson. Over the past 18 months, Three UK has collaborated with Ericsson to enhance network energy performance using industry-leading energy-efficient radios and advanced AI and data analytics.

In late 2023, Three UK became one of the first major UK operators to deploy Ericsson’s award-winning dual-band Radio 4490, which is 25% lighter and consumes less power than previous models. This technology simplifies site access and expedites upgrades. Additionally, Three UK has implemented software features that reduce power consumption during low traffic hours, leveraging advanced machine learning, passive cooling, and power-saving functionalities. These radios autonomously switch off components when inactive and reactivate them in microseconds for service requests.

Key Points:

  • Three UK deploys Ericsson’s energy-efficient dual-band Radio 4490, reducing power consumption and site weight.
  • Software features implemented to lower power use during low traffic hours with AI and machine learning.
  • Advanced radios autonomously manage 4G and 5G networks, enhancing efficiency.
  • Partnership with Ericsson improves network energy efficiency by up to 70% at selected sites, while reducing CO2 emissions.

undefined

ST Telemedia Global Data Centres raises US$1.3B from KKR and Singtel

ST Telemedia Global Data Centres (STT GDC), a leading data center colocation services provider, announced a significant investment from a consortium led by KKR and Singtel. The consortium will invest SGD 1.75 billion (~USD 1.3 billion) initially through Redeemable Preference Shares (RPS) with detachable warrants. This investment represents the largest digital infrastructure deal in Southeast Asia in 2024. Upon full exercise of the warrants, the consortium will contribute an additional SGD 1.24 billion (~USD 920 million).

The investment follows an independent competitive selection process by STT and STT GDC, focusing on the consortium’s expertise, financial strength, and strategic vision. The funds will support STT GDC's ongoing international expansion and growth plans through both organic and inorganic strategies. Post-transaction, ST Telemedia will retain its position as the majority shareholder of STT GDC.

STT GDC, headquartered in Singapore, is among the world's fastest-growing data center providers, operating over 95 data centers across 11 countries and serving more than 20 major business markets. With a combined capacity of over 1.7GW of IT load, STT GDC offers high-quality colocation, connectivity, and round-the-clock support services, solidifying its critical role in the digital infrastructure landscape.

Key Points:

  • Investment Details: SGD 1.75 billion (~USD 1.3 billion) initially, with a potential additional SGD 1.24 billion (~USD 920 million) from the KKR-led consortium.
  • Selection Process: Chosen for their expertise, financial strength, and strategic vision.
  • STT GDC Profile: Headquartered in Singapore, with 95+ data centers in 11 countries and over 1.7GW of IT load capacity.
  • Strategic Use: Funds to support international expansion and growth, maintaining ST Telemedia as the majority shareholder.
  • Industry Impact: Largest digital infrastructure investment in Southeast Asia for 2024.

https://www.singtel.com/about-us/media-centre/news-releases/st-telemedia-global-data-centres-raises-1-75-billion

undefined

onsemi plans silicon carbide fab in Czech Republic

onsemi announced plans to establish a state-of-the-art, vertically integrated silicon carbide (SiC) manufacturing facility in the Czech Republic to produce intelligent power semiconductors for use in electric vehicles, renewable energy and AI data centers.

Silicon carbide is a critical material for high-power, high-temperature applications, and is extremely difficult to produce. onsemi is one of the only companies in the world with the ability to manufacture SiC-based semiconductors from crystal growth to advanced packaging solutions.

onsemi’s plan to expand SiC manufacturing with a multi-year brownfield investment of up to $2 billion (44 billion CZK) is part of the company’s previously disclosed long-term capital expenditure target. This investment would build on the company’s current operations in the Czech Republic, which include silicon crystal growth, silicon and silicon carbide wafer manufacturing (polished and EPI) and a silicon wafer fab. Today, the site can produce more than three million wafers annually, including more than one billion power devices. Upon completion, the operation would contribute annually more than $270 million USD (6 billion CZK) to the country’s GDP.

"Our brownfield investment would establish a Central European supply chain to better service our customers’ rapidly increasing demand for innovative technologies that improve the energy efficiency in their applications,” said Hassane El-Khoury, president and CEO, onsemi. “Through a close collaboration with the Czech government, the expansion would also enhance our production of intelligent power semiconductors that are essential to helping ensure the European Union is able to achieve its ambitions to significantly reduce carbon emissions and environmental impact.”


undefined

Tuesday, June 18, 2024

HPE teams with NVIDIA on Private-Cloud AI 

 HPE and NVIDIA have co-developed the Private Cloud AI portfolio to enhance innovation, return on investment, and manage enterprise AI risks. This portfolio supports a range of AI workloads and offers robust enterprise controls, cloud experiences, and flexible consumption models. The curated AI and data software stack is built on NVIDIA AI Enterprise and includes HPE AI Essentials software for adaptable and compliant AI solutions. The portfolio integrates NVIDIA's advanced computing technologies and HPE's infrastructure and cloud services for optimal performance and manageability across hybrid environments.

The announcement, which occured during the HPE Discover keynote by HPE President and CEO Antonio Neri, who was joined by NVIDIA founder and CEO Jensen Huang, marks the the expansion of a long-running partnership between the firms.

HPE Private Cloud AI

  • Inference, Fine-Tuning, and RAG AI Workloads: Supports proprietary data utilization.
  • Enterprise Control: Ensures data privacy, security, transparency, and governance.
  • Cloud Experience: Provides ITOps and AIOps capabilities for increased productivity.
  • Flexible Consumption: Adapts to future AI opportunities and growth.

Curated AI and Data Software Stack

  • NVIDIA AI Enterprise: Accelerates data science pipelines and streamlines AI application development.
  • NVIDIA NIM Microservices: Optimizes AI model inferencing for smooth transitions from prototype to deployment.
  • HPE AI Essentials: Offers ready-to-run AI and data foundation tools with a unified control plane for adaptable solutions and enterprise support.

Integrated AI Infrastructure Stack

  • NVIDIA Spectrum-X Ethernet Networking
  • HPE GreenLake for File Storage
  • HPE ProLiant Servers: Supports NVIDIA L40S, H100 NVL Tensor Core GPUs, and GH200 NVL2 platform.

Global System Integrator Collaboration

  • Accelerated AI Solutions: Deloitte, HCLTech, Infosys, TCS, and WIPRO support the NVIDIA AI Computing by HPE portfolio for industry-focused AI solutions.

Support for NVIDIA’s Latest GPUs and CPUs

  • HPE Cray XD670: Supports eight NVIDIA H200 NVL Tensor Core GPUs for LLM builders.
  • HPE ProLiant DL384 Gen12: Equipped with NVIDIA GH200 NVL2 for larger models and RAG consumers.
  • HPE ProLiant DL380a Gen12: Supports up to eight NVIDIA H200 NVL Tensor Core GPUs for flexible scaling of GenAI workloads.
  • Future Support: Includes NVIDIA GB200 NVL72 / NVL2 and new NVIDIA architectures (Blackwell, Rubin, Vera).

High-Density File Storage

  • NVIDIA DGX BasePOD Certification: Provides enterprise file storage solutions for AI and GPU-intensive workloads.
  • Upcoming Certifications: HPE will partner on new NVIDIA reference architecture storage certifications.

Availability

  • HPE Private Cloud AI: Expected in the fall.
  • HPE ProLiant DL380a Gen12 and DL384 Gen12 Servers: Available in the fall.
  • HPE Cray XD670 Server: Available in the summer.
undefined

Vantage Data Centers completes $9.2B equity investment

Vantage Data Centers completed a $9.2 billion equity investment led by investment vehicles managed by DigitalBridge Group and Silver Lake.

Vantage has grown in recent years to own or control more than 25 sites in North America and EMEA totaling over three gigawatts of expected capacity. As part of the company’s investment plan, the new funding is expected to accelerate further growth by driving an estimated $30 billion of additional development, extending Vantage’s track record of bringing market leading capacity and innovative solutions to customers around the world. In connection with the investment, Vantage will continue its development of next-generation data centers, including energy-efficient and sustainable designs purpose-built for AI and large-scale cloud deployments.

“It’s exciting to experience such strong investor demand to fuel our unprecedented growth across North America and EMEA,” said Sureel Choksi, president and CEO of Vantage. “The confidence that DigitalBridge, Silver Lake and other investors have in Vantage and our ability to execute, as evidenced by an investment round that was significantly oversubscribed, is a testament to our team’s track record of delivering for our customers. This new funding from the world’s leading digital infrastructure and technology investors, along with numerous co-investors, is a game changer that uniquely positions Vantage to capitalize on the incredible AI and cloud opportunity in front of us.”

The capital raise was significantly oversubscribed and upsized by $2.8 billion with participation from multiple global investors, bringing the completed investment to $9.2 billion. Together with the €1.5 billion investment by AustralianSuper announced in September 2023, aggregate new investment in Vantage over the past nine months totals approximately $11 billion, of which more than $7 billion is primary equity. Primary proceeds will be used to fund Vantage’s growth across North America and EMEA, accelerating and extending the company’s strategic capabilities to partner with global hyperscalers in meeting unprecedented cloud and AI demand.


Microsoft opens data center region in Madrid

Microsoft inaugurated its first cloud region of data centers in Spain, located in multiple locations in the Community de Madrid.

The new cloud region (called Spain Central) will help to accelerate the digitalization of Spanish organizations, the development of the AI Economy in Spain and the growth of the local IT industry. The technologies offered at Spain Central are, among others, the most modern services for productivity, data analytics, cybersecurity, computing and storage (Azure, Microsoft 365, Dynamics 365, Power Platform), and also include generative Artificial Intelligence services, thus placing this cloud region among the most advanced in Europe. The new cloud region will provide highly available, scalable, and secure cloud services, including the option of in-country data residency for Spanish organizations.

Microsoft expects to invest US$2.1 billion in the period 2024-2025 in Spain, responding to the growing demand for Microsoft’s AI and cloud solutions from Spanish organizations.

Customers already using the new cloud region include (among others): Abanca, Aciturri, Banc Sabadell, Banco Santander, Bankinter, BBVA, CaixaBank, Canal de Isabel II, CELSA, Codere, Comunidad de Madrid, Grupo Catalana Occidente, Ibercaja, Ilunion, Metrovacesa, Mutua Madrileña, Naturgy, Navantia, Pelayo, Prosegur, Repsol, Ribera Salud, SegurCaixa Adeslas, Tecnicas Reunidas and Velilla. In addition, software developers (ISVs) such as Barracuda, Citrix, NetApp, OpenAI, Palantir, SAP, Veeam, Vmware or Zertia are also making use of Microsoft’s new infrastructure in

Padtec supplies subsea network connect Belém to Fortaleza for Global Fiber

Global Fiber, a telecommunications service operator specialized in optical fiber operations in Peru and Brazil, has chosen Padtec and Satelital as the primary suppliers for the submarine solution for a festoon network connecting two Brazilian capitals: Belém (in Pará state) and Fortaleza (in Ceará state). A festoon network is a type of underwater infrastructure that links multiple coastal points without requiring intermediate amplification.

Global Fiber operates a fiber optic network with coverage throughout Peru and is implementing a sub-river network on the Amazon River, connecting the locations of Iquitos and Santa Rosa de Yaraví, in Peru, Leticia, in Colombia, and Tabatinga, in Brazil. “By supporting the construction of the Belém-Fortaleza section, within the Lima – Fortaleza project, Global Fiber contributes to reducing digital gaps in difficult-to-access locations, providing all areas of the Amazon region with high-quality communication and a more connected Latin America”, highlights Obed Dionisio, CEO of Global Fiber.

Padtec and Satelital will be responsible for leading the construction of the first Global Fiber festoon network in Brazil. Manuel Andrade, CEO of Padtec North America and director of International Operations, states that Padtec is returning to the submarine and sub-river systems market. “With the Belém – Fortaleza section, which is expected to come into operation at the end of 2025, Padtec expands its experience in implementing submarine networks with these characteristics”, he states. 


https://www.padtec.com.br/en/padtec-will-provide-a-solution-for-the-submarine-network-of-global-fiber-that-connects-belem-to-fortaleza/

undefined

Dell'Oro: Accelerated Servers accounted for over 50% of server sales

Accelerated servers equipped with GPUs and custom accelerators accounted for more than half of all server sales in 1Q 2024, according to a new report from Dell'Oro Group. Not only are the hyperscalers leading the way in deploying accelerated servers for AI workloads, but enterprises are also increasing their adoption of accelerated servers as GPU supplies improve.

“Worldwide data center capex experienced the first double-digit growth in five quarters. The hyperscale Cloud service providers in US and China saw an aggregate growth rate exceeding 40 percent as AI-infrastructure investments maintained strong growth momentum,” said Baron Fung, Sr. Research Director at Dell’Oro Group. “Meanwhile, there is optimism for a modest improvement in the outlook for enterprise IT spending along with a recovery in general-purpose server market later this year as end users replace servers overdue for refresh to the latest CPU platforms from Intel and AMD,” explained Fung.

Additional highlights from the 1Q 2024 Data Center IT Capex Quarterly Report:

  • Server and storage system revenues are forecast for growth greater than 30 percent in 2024, while network and physical infrastructure revenues are expected to grow single digits.
  • Dell led all OEMs in server revenue in 1Q 2024 followed by HPE and Supermicro. Accelerated servers accounted for an estimated 27 percent of the OEM server revenues during the quarter.
  • White box server vendors gained 11 points of revenue share year-over-year compared to the OEMs as hyperscalers increased deployments of servers equipped with GPUs and custom accelerators.
  • The accelerated server market is forecast to increase by more than 80 percent in 2024.

undefined

Monday, June 17, 2024

Kyocera's ceramic mirrors selected fo ISS-Earth optical comms

Kyocera's Fine Cordierite ceramic mirrors are set to play a crucial role in advancing optical communication between the International Space Station (ISS) and Earth. These mirrors have been incorporated into the Quantum-Small Optical Link (QSOL) developed by Sony Computer Science Laboratories. Commissioned by Japan's Ministry of Internal Affairs and Communications, QSOL is a key component of the Secure Laser Communications Terminal for Low Earth Orbit (SeCRETS) project, aimed at demonstrating advanced on-orbit technology.

The project is a collaborative effort involving the National Institute of Information and Communications Technology, the Next Generation Space System Technology Research Association, SKY Perfect JSAT, and Sony CSL. This groundbreaking experiment focuses on overcoming the current limitations of data transmission between Earth observation satellites and ground stations. Traditionally, communication relies on optical wireless methods using radio waves or visible light, which are increasingly inadequate due to the growing volume of data generated by advanced satellite sensors.

The introduction of Kyocera's Fine Cordierite ceramic mirrors is expected to revolutionize this field. These mirrors offer exceptional thermal and mechanical properties, such as low thermal expansion and long-term dimensional stability, crucial for maintaining the precise alignment needed for high-speed, high-capacity data transmission. The successful implementation of this technology could lead to data transmission speeds over 100 times faster than current radio wave methods, significantly enhancing the capabilities of space infrastructure.


Key Points of the Experiment

Kyocera's Contribution:

Fine Cordierite ceramic mirrors supplied for ISS-Earth optical communication.

QSOL Development:

Created by Sony Computer Science Laboratories under the Ministry of Internal Affairs and Communications.

Collaborative Effort:

Involves the National Institute of Information and Communications Technology, Next Generation Space System Technology Research Association, SKY Perfect JSAT, and Sony CSL.

Current Communication Methods:

Utilizes optical wireless communication with radio waves or visible light, essential for various data-dependent applications.

Challenges and Solutions:

Addressing the need for rapid transmission of large data volumes from Earth observation satellites.

Advantages of Kyocera’s Fine Cordierite Ceramic Mirrors:

Low Thermal Expansion: Ensures minimal dimensional changes, crucial for nanoscale precision.

High Mechanical Strength and Rigidity: Offers significantly greater strength and rigidity compared to low thermal expansion glass.

Long-Term Dimensional Stability: Provides stable performance over extended periods.

Radiation Resistance: Maintains consistent performance under radiation, suitable for space environments.

SeCRETS Launch and Experiment:

Launched on August 2, 2023, and installed on the ISS's Kibo module.

Demonstrated secure optical communication using one-time pad encryption between the ISS and a ground station.


https://www.sonycsl.co.jp/press/prs20240418/
https://www.sonycsl.co.jp/press/prs20240423/?lang=en

false

Equinix to export heat from Paris data centre to Olympic Aquatic Centre

 Starting next month, Equinix will launch a significant Heat Export project at its newest data center in Paris, PA10. The recovered heat will be transferred to the Plaine Saulnier urban development zone and the Olympic Aquatic Centre, venues for the 2024 Summer Olympics. This initiative is set to provide sustainable heating solutions for the Olympic facilities, showcasing a practical application of green energy technology on a global stage.


Equinix is calling for municipal planning agencies, energy utilities, and heat network operators worldwide to join its innovative Heat Export program, which aims to harness the residual heat generated in its data centers for community use. This program offers a cleaner alternative to traditional heating methods, crucial during a time of rising global energy prices and supply constraints.

The Heat Export program recovers residual heat from Equinix's data centers and distributes it to nearby buildings, helping to reduce the reliance on conventional energy sources. Equinix has successfully implemented several Heat Export initiatives across Europe and the Americas, with projects in France, Finland, Germany, Ireland, Switzerland, and Canada.

In Toronto, Equinix's TR5 IBX data center already distributes recovered heat to multiple residential buildings, a hotel, a university, and a shopping center in downtown Markham. Additionally, TR5 supplies domestic hot water year-round to several buildings through a partnership with Markham District Energy.

Equinix said it is dedicated to expanding its Heat Export program in suitable locations, identifying data centers that can serve as valuable heat sources for local communities. The company is actively seeking partnerships with heat network operators, energy utilities, and local municipalities to further these projects. By collaborating with these entities, Equinix aims to make a significant positive impact on the communities it serves.

"Our Heat Export program is one important way data centers can give back to their local communities," explained Gary Aitkenhead, SVP, EMEA IBX Operations, Equinix. "However, it is only possible through collaboration with energy providers, governments, and other third-party enablers. So, we are asking for more partners to join the program and re-use the heat from our operations for the good of people and our planet."

false

Infinera hosts demo of 5G xHaul using coherent Open XR optics

Infinera hosted a multi-vendor demonstration of 5G mobile broadband xHaul using coherent open XR optics point-to-multipoint optical transmission.

Hosted at the European Open Test & Integration Center in Torino by TIM, the high-capacity xHaul application testing included fronthaul, midhaul, and backhaul transport scenarios with XR-based coherent pluggable optics deployed in third-party hosts supporting point-to-point and point-to-multipoint optical transmission. The results of the performance testing included a successful demonstration of xHaul synchronization and timing distribution in a point-to-multipoint optical transport architecture.

Equipment used in the interoperable xHaul testing included Infinera ICE-X intelligent coherent pluggables, the DZS Saber 2200, and Calnex Paragon-NEO. Part of the work carried out by TIM and Infinera was supported by the EU project ALLEGRO, GA No. 101092766.

Infinera says the demo represents a key step toward enabling mobile operators to greatly simplify and reduce the cost of 5G and next-generation mobile transport network rollouts through the reduction of the number of optical transceivers, resulting in significant total cost of ownership savings.

“It is not only the significant bandwidth demands of 5G that create challenges for mobile operators, but also the fundamental misalignment between actual 5G network traffic patterns and the underlying transport technology,” said Ron Johnson, SVP and General Manager, Optical Subsystems and Global Engineering Group, Infinera. “Working in close collaboration with industry-leading mobile operators such as TIM, this testing validates the critical role that XR optics innovation can play in transforming the economics of 5G transport and paving the way for efficient 6G networks.”

false

Aim raises $18 million for AI enterprise security

Aim Security, a start-up based in Tel Aviv, has announced $18 million in Series A funding for its development of AI-enabled enterprise security solutions.

Aim’s proactive, all-encompassing security platform governs and secures all forms of AI use across the modern business environment. Specifically tailored for AI’s unique threats, including sensitive data exposure, supply chain vulnerabilities, harmful or manipulated outputs, and the emergence of attack methods such as jailbreaks and prompt injection, Aim’s platform has already engendered significant customer trust across highly regulated industries, including banking, insurance, healthcare, manufacturing, and defense.

The funding round was led by Canaan Partners, with participation from the company’s seed investor, YL Ventures. Aim’s backers include the founders of WIZ and industry leaders from Palo Alto Networks, Proofpoint, and Google. This is among the fastest Series A rounds in cybersecurity history and brings Aim’s total funding to $28 million.

Aim Security was founded by cybersecurity veterans Matan Getz, CEO, and Adir Gruss, CTO, who have broken and built AI platforms, including in the IDF’s elite intelligence Unit 8200.

“Enterprise customers from a wide range of industries repeatedly asked us, ‘How can we embrace the undeniable advantages of cutting-edge AI technologies, such as heightened efficiency and optimized decision-making, without crossing our acceptable risk threshold?’” said Matan Getz, CEO of Aim Security. “At Aim, our mission is clear: to serve as the trusted AI security ally for enterprise security leaders and allow organizations to confidently unleash the potential of these technologies within their environment, knowing that Aim is their protection layer.”

false

Sunday, June 16, 2024

2Africa subsea cable reaches UK

The 2Africa subsea cable has been landed in Bude, Cornwall.

At 45,000km, 2Africa is expected to be the longest subsea cable ever deployed. A total of 46 cable landing locations are envisioned for the project before completion in 2024. With a capacity of up to 180 terabits per second (Tbps), 2Africa will be the first system to provide continuous capacity around the African continent, and will connect 33 countries on three continents (Europe, Africa and Asia). 

Vodafone, which is the landing partner for the the 2Africa subsea cable, in Bude, confirmed that it is now carrying traffic on part of its 2Africa cable system to deliver additional capacity to customers.

As the lead partner for the UK landing, Vodafone handled the physical cable coming ashore; burial on the beach; and installation into a new, purpose-built beach manhole, where 2Africa will connect to terrestrial cable routing back to the existing Vodafone Cable Landing Station. Vodafone will also provide 2Africa with onward connectivity via two diverse terrestrial infrastructure and fibre routes linking to the London area.

Vodafone has named the twin-fibre pairs it owns in the cable system after Vodafone Group’s Head of Subsea Partnerships, Rick Perry. The Vodafone system is referred to as SHARP (the System Honouring the Achievements of Rick Perry).

The 2Africa consortium is made up of eight international partners: Bayobab; Center3; China Mobile International; Meta; Orange; Telecom Egypt; Vodafone Group; and WIOCC. Alcatel Submarine Networks is responsible for the manufacture and installation of the 2Africa cable.

Vodafone has been the partner for 11 landings of the 2Africa cable to date, with its experience in subsea cables tracing back to the landing of the first-ever transoceanic cable in Porthcurno, Cornwall, in 1870.

YOFC and China Mobile test 800G hollow-core fiber

Yangtze Optical Fibre and Cable YOFC), in collaboration with China Mobile, unveiled the world's first 800G hollow-core fibre transmission test network in Shenzhen-Dongguan, Guangdong. Hollow-core fibre, utilizing an innovative air-guiding mechanism, surpasses the performance constraints associated with traditional solid-core fibres in both capacity and latency. 

YOFC's participation in the test network featured a comprehensive suite of solutions specifically designed for hollow-core fibre optics, encompassing the provision of cables, essential splicing services, and adapters for integration with single-mode fibres. The project successfully navigated several technical challenges common in complex installations. The challenges involved mitigating environmental risks and ensuring robust infrastructure during fibre laying and splicing operations—addressing pipe flooding, managing excessive humidity, and conducting precise splicing in outdoor environments. These efforts were crucial in enabling a detailed evaluation of the fibre's performance under real operational conditions.

Key performance metrics achieved on a 20km test link—such as splice losses between hollow-core fibres, connections from hollow-core to solid-core fibres, and attenuation post-installation—were benchmarked against leading global standards. Moreover, the deployment featured an 800Gbps single-wavelength optical transmission setup, achieving a bidirectional long-distance transmission test with a throughput of 128Tb/s over hollow-core fibres. This not only demonstrated the practical application of such fibres in intricate and demanding network environments but also served as a critical proof-of-concept for the scalability and industrial applicability of hollow-core fibre technologies.