<div>
  <ul>
    <li>
      <p>
        The US-China Economic and Security Review Commission (USCC) has proposed a Manhattan Project-style initiative to develop artificial general intelligence (AGI) systems, emphasizing public-private partnerships amid growing technological competition with China.
      </p>
    </li>
    <li>
      <p>
        The Commission suggests streamlining data center permitting processes to accelerate AI development, noting that energy infrastructure is a significant bottleneck for training large AI models.
      </p>
    </li>
    <li>
      <p>
        The USCC report also recommends eliminating the &quot;de minimis&quot; trade exemption, which allows Chinese goods under USD 800 to enter the US with minimal inspection. It cites concerns over the influx of 1.4 billion shipments annually.
      </p>
    </li>
    <li>
      <p>
        The Commission, established by Congress in 2000, provides annual recommendations on US-China relations and aims to guide lawmakers on economic and strategic competition issues with China.
      </p>
    </li>
  </ul>
  <p>
  </p>
</div>


<div>
  <p>
    <a href="https://www.reuters.com/technology/artificial-intelligence/us-government-commission-pushes-manhattan-project-style-ai-initiative-2024-11-19/">
      Reuters
    </a>
  </p>
</div>


Machine Learning Infrastructure

PROMINENT_CATEGORY

<div>
 <p>The US-China Economic and Security Review Commission (USCC) has advised a Manhattan Project-style initiative to develop artificial general intelligence (AGI) systems, highlighting the role of public-private partnerships in escalating technological rivalry with China.</p>
 <p>To expedite AI development, the Commission recommends easing the processes related to data center permissions, identifying power infrastructure as a crucial hurdle for training expansive AI models.</p>
 <p>The "de minimis" trade exemption, allowing Chinese goods of less than USD 800 to get into the US with limited inspection, should be abolished according to the USCC report. This is in response to concerns over 1.4 billion shipments entering annually.</p>
 <p>Congress established the Commission in 2000; it provides yearly recommendations on US-China relations, with a goal of steering lawmakers on China-related economic and strategic competition matters.</p>
</div>

US Commission proposes Manhattan Project-style AI development initiative

<div>
 <p>SuperAnnotate, a GenAI dataset management platform, has secured USD 36 million in Series B funding. The round was led by Socium Ventures and saw participation from NVIDIA, Databricks Ventures, Play Time Ventures, Glynn Capital, among others. So far, the company's total funding has reached USD 53 million.</p>
 <p>The raised capital is intended to be used to increase the number of employees, which is currently 100, to invest in product research and development, and to expand its customer base. Currently, they serve over 100 companies.</p>
 <p>The platform provided by SuperAnnotate assists users in creating and monitoring GenAI training datasets. It offers tools for fine-tuning, iterating, and evaluating these datasets. Among their prestigious clients are companies like Databricks and Canva. Additionally, the company also provides access to a marketplace of crowd-sourced workers for data annotation tasks.</p>
</div>

SuperAnnotate raises USD 36 million in Series B funding for R&D and headcount expansion

ANALYST_QUICK_TAKE

INCUMBENTS_PRESENT

<div>
 <p>SAS, the software firm, has bought the main software properties of Hazy, a company from London that leads in synthetic data technology. The price of the acquisition was not disclosed.</p>
 <p>This purchase will improve SAS' Data Maker platform, enabling users to simulate a range of possible future circumstances. The benefits include a quicker route to market for AI initiatives through the use of synthetic datasets, enhanced data privacy, and reduced costs in data collection. These improved solutions are anticipated to be ready for preview in early 2025.</p>
 <p>Hazy specializes in synthetic data technology which replicates the statistical trends of real data while avoiding the exposure of private, identifiable, or protected information. This aids businesses in tackling data scarcity and privacy problems, while enhancing the accuracy of models.</p> 
</div> <p><b>Analyst QuickTake:<br>
 </b> The news aligns with SAS's USD 1 billion commitment in <a href="https:<br>
//itbrief.co.uk/story/sas-invests-1-billion-to-propel-generative-ai-in-multiple-sectors-expands-microsoft-partnership"> September 2023 </a> to expand GenAI solutions across multiple industries, including banking, life sciences, and retail. The M&amp;A also expands on the company's existing machine learning operations, including <a href="https:<br>
//www.sas.com/en_hk/solutions/ai/generative-ai.html"> SAS Viya </a> , an LLM deployment solution.</p>

SAS acquires Hazy to leverage synthetic data generation capabilities

<div>
 <p>AMD has unveiled the Versal Premium Series Gen 2, a cutting-edge platform designed for system-on-chip applications. The platform dynamically adapts like a field programmable gate array (FPGA) and is tailored for data-intensive AI work in data centers, including those in the aerospace and defense niche. It will be developed and tested, with development tools available from Q2 2025 and silicon samples in early 2026. Production shipments are expected to begin in 2H 2026.</p>
 <p>Key features of the chip include a Compute Express Link (CXL) 3.1, PCIe Gen6, and LPDDR5X memory support in hard IP. The chip facilitates high-bandwidth interconnectivity between processors and accelerators and boasts ultra-fast LPDDR5X memory speed of up to 8533 Mb/s. The platform incorporates PCIe Integrity and Data Encryption in hard IP, ensuring high security and integrity of data.</p>
 <p>AMD asserts that the new platform surpasses its competitors possessing PCIe Gen4 or Gen5 support with 2x–4x faster line rates. It also assures up to 2.7 times faster host connectivity over its rivals using LPDDR4/5 memory. Furthermore, the platform permits scalable memory pooling and extension for multiple accelerators, contributing to a more efficient and effective computation environment.</p>
</div> <p><b>Analyst QuickTake:<br>
 </b> AMD launched the Versal Premium VP1902 adaptive system-on-chip (SoC) in <a href="https:<br>
//timesofindia.indiatimes.com/gadgets-news/amd-announces-versal-premium-vp1902-soc-all-the-details/articleshow/101316199.cms"> June 2023 </a> . This SoC is noted for being the world’s largest adaptive SoC at the time, designed to enhance the verification of complex semiconductor designs and improve performance in various applications, including AI and autonomous technologies.</p>

AMD to ship Versal Premium Series Gen 2 adaptive chip platform in 2H 2026

<div>
 <p>British AI chip networking startup Oriole Networks has attained USD 22 million in a Series A funding round dominated by Plural. Current investors like UCL Technology Fund, XTX Ventures, Clean Growth Fund, and Dorilton Ventures also participated.</p>
 <p>This influx of capital will enable Oriole Networks to expand its team, collaborate with large-scale suppliers, and propel its growth. By 2025, the company aims to deliver its novel products to customers, fostering a photonic networking environment specifically for AI applications.</p>
 <p>Established in 2023, the company's mission is creating a "super brain" by connecting thousands of AI chips using light, which is key to training LLMs. Oriole Networks claims its technique hastens the LLM training process by a hundred folds compared to traditional means and significantly reduces energy consumption. It aspires to provide sustainable computing solutions to cater to the ever-increasing demands of data centers. </p>
</div>

Oriole Networks raises USD 22 million in Series A funding to expand team and engage with suppliers

<div>
 <p>Taiwan Semiconductor Manufacturing Company (TSMC) plans to cease production of AI chips at seven nanometers or smaller for Chinese customers. Future supplies will be subject to Washington's approval.</p>
 <p>Major Chinese tech firms like Alibaba and Baidu, who depend on TSMC for advanced AI chips, will be affected. Specifically, Baidu's Kunlun II processor is largely impacted as it is produced using TSMC's 7nm technology.</p>
 <p>This move is a result of an ongoing US Commerce Department investigation into Huawei AI devices that contain TSMC-produced chips and the expected new US restrictions on chip exports to China.</p>
 <p>The company mentions these restrictions are aimed primarily at AI processors while the larger application remains unclear. However, TSMC assures that these new constraints will not have a significant effect on its revenue, as seen by a 29.2% revenue increase in October, amounting to approximately USD 9.8 billion.</p>
 
</div> <p><b>Analyst QuickTake:<br>
 </b> TSMC's decision comes amid US export controls on AI chips to China, initially implemented in <a href="https:<br>
//thediplomat.com/2024/04/ai-chips-for-china-face-additional-us-restrictions/"> October 2022 </a> . These rules were further tightened in <a href="https:<br>
//www.reuters.com/technology/us-commerce-updates-export-curbs-ai-chips-china-2024-03-29/"> March 2024 </a> , clarifying that limitations also apply to laptops containing advanced AI chips.</p>

TSMC halts 7nm chip production for Chinese customers

<div>
 <p>Hardware provider Graphcore is expanding its workforce by 20%, creating 75 new roles globally. The jobs will be in various fields such as silicon design, hardware, datacenter infrastructure, software, and AI research. Opportunities are available in Bristol, Cambridge, London, Gdańsk (Poland), and Hsinchu City (Taiwan).</p>
</div> <p><b>Analyst QuickTake:<br>
 </b> Last year, Meta <a href="https:<br>
//www.reuters.com/technology/meta-platforms-scoops-up-ai-networking-chip-team-graphcore-2023-05-05/"> recruited </a> Graphcore's team of AI networking technology experts to bolster its data center initiatives. At the time, Graphcore was grappling with significant financial challenges, including a pre-tax loss of USD 205 million in <a href="https:<br>
//www.reuters.com/technology/losses-widen-cash-needed-chip-startup-graphcore-an-nvidia-rival-filing-2023-10-05/"> 2022 </a> due to a 46% YoY revenue decline. Additionally, the company faced setbacks from US export regulations, leading to substantial layoffs and the cessation of its operations in China by <a href="https:<br>
//www.bloomberg.com/news/articles/2023-11-22/nvidia-rival-graphcore-pulls-out-of-china-citing-export-rules"> November 2023 </a> . However, following its acquisition by SoftBank in <a href="https:<br>
//techcrunch.com/2024/07/11/softbank-acquires-uk-ai-chipmaker-graphcore/"> July 2024 </a> , the company has signaled a comeback, marked by the 20% workforce increase.</p>

Graphcore to expand workforce by 20% to help improve its GenAI compute offerings

<div>
 <p>Saudi Arabia's Public Investment Fund (PIF) is reportedly contemplating the establishment of a new Artificial Intelligence (AI) division known as Project Transcendence. The government-backed initiative potentially has a funding pool of up to USD 100 billion.</p>
 <p>Modeled after PIF's Alat electronics group, the planned project will concentrate primarily on investing in data centers, startups, and infrastructure. The main goal is to facilitate the development of AI products and solutions. In addition to this, Project Transcendence envisages forging partnerships with high-profile tech companies on a global scale.</p>
 <p>One of the objectives of Project Transcendence is the amalgamation of various government entities. This is geared towards stimulating job creation for Saudi citizens and nurturing domestic talent in AI and its associated technologies.</p>
 <p>This scheme is distinctly separate from the reported negotiations that PIF is partaking in with Andreessen Horowitz and other reputed venture capital firms. The dialogue revolves around constituting a hefty AI-investment fund that is estimated to be worth USD 40 billion.</p>
</div>

Saudi Arabia plans USD 100 billion AI infrastructure project

<div>
 <p>Advex AI, a synthetic data generator, has secured a seed funding amount of USD 3.5 million. The funding round was led by Construct Capital and included contributions from Pear VC, Emerson Collective, and angel investors Arash Ferdowsi and Ankit Jain.</p>
 <p>This fund will be channeled towards addressing the increasing demand from global manufacturers. The company is also in the process of recruitment for various positions to support its growth.</p>
 <p>Utilizing GenAI, Advex creates synthetic data that is then used in the training of computer vision models. The company specializes in providing solutions for the logistics sector. Its platform generates images for defect detection, spotting issues such as knots and resin.</p>
 
</div> <p><b>Analyst QuickTake:<br>
</b> The boom in the GenAI industry has increased the demand for high-quality datasets used in model training. Recent funding rounds to meet this demand include <a href="https:<br>
//sp-edge.com/updates/34944"> Mindtech </a> 's ~USD 5.7 million raise and <a href="https:<br>
//www.eu-startups.com/2023/10/trieste-based-aindo-bags-e6-million-series-a-to-use-synthetic-data-for-training-machine-learning-models/"> Aindo </a> 's ~USD 6.5 million round.</p>

Advex AI raises USD 3.5 million in seed funding to develop AI vision platform

<div>
 <p>GMI Cloud, a company that provides AI cloud infrastructure, has successfully raised USD 82 million in Series A financing. The leading contributors in this round of funding were Headline Asia, Banpu Next, and Wistron.</p>
 <p>The company will utilize these funds to establish a new data center in Colorado and enlarge their GPU footprint across the US. In addition, they aim to enhance their platform's capabilities and expand their global AI infrastructure. GMI Cloud, already having strong footprints in Taiwan and other regions in Asia-Pacific, sees this as an opportunity for further growth.</p>
 <p>The GMI Cloud platform is powered by NVIDIA's H200 GPUs and has a unique Cluster Engine designed for cloud management. The company’s solution includes features like multi-tenant private clouds and Kubernetes orchestration for managing resources, with an emphasis on performance, scalability, reliability, and security.</p>
</div> <p><b>Analyst QuickTake:<br>
</b> GMI Cloud's funding round underscores the increasing demand for infrastructure to train and enhance the inference capabilities of foundation models. This trend aligns with recent investments in photonics startups, which aim to reduce power consumption and boost FM performance. Notable examples include <a href="https:<br>
//sp-edge.com/updates/34972"> LightMatter’s </a> USD 400 million round and <a href="https:<br>
//www.oriolenetworks.com/news1/series-a-press-release"> Oriole Network’s </a> USD 22 million round.</p>

All Updates

US Commission proposes Manhattan Project-style AI development initiative

SuperAnnotate raises USD 36 million in Series B funding for R&D and headcount expansion

SAS acquires Hazy to leverage synthetic data generation capabilities

AMD to ship Versal Premium Series Gen 2 adaptive chip platform in 2H 2026

Oriole Networks raises USD 22 million in Series A funding to expand team and engage with suppliers

TSMC halts 7nm chip production for Chinese customers

Graphcore to expand workforce by 20% to help improve its GenAI compute offerings

Saudi Arabia plans USD 100 billion AI infrastructure project

Advex AI raises USD 3.5 million in seed funding to develop AI vision platform

GMI Cloud raises USD 82 million in Series A funding to expand AI infrastructure

Machine Learning Infrastructure

US Commission proposes Manhattan Project-style AI development initiative

Contact us