<div>
  <p>
  </p>
  <ul>
    <li>
      <p>
        Voxel51, a visual AI solution provider, has secured USD 30 million in a Series B financing round led by Bessemer Venture Partners, with participation from Tru Arrow Partners, Drive Capital, Top Harvest Capital, and Shasta Ventures, among others.
      </p>
    </li>
    <li>
      <p>
        Voxel51 plans to deploy this fresh capital to accelerate its product roadmap and go-to-market initiatives. The company also aims to boost AI research and development.
      </p>
    </li>
    <li>
      <p>
        Formed in 2018, Voxel51 is an AI software company that enables machine learning and computer vision scientists to rapidly curate and experiment with their visual datasets.
      </p>
    </li>
  </ul>
  <p>
  </p>
</div>


<div>
  <p>
    <a href="https://www.prnewswire.com/news-releases/voxel51-raises-30m-series-b-funding-to-make-visual-ai-a-reality-302147665.html">
      Press release
    </a>
  </p>
</div>


Machine Learning Infrastructure

<div>
 
 <p>Voxel51, a provider of AI visual solutions, has garnered USD 30 million in a Series B funding round. Bessemer Venture Partners was the lead investor, with contributions also coming from Tru Arrow Partners, Drive Capital, Top Harvest Capital, and Shasta Ventures, among others.</p>
 <p>The firm intends to utilise the newly-acquired capital to expedite its product roadmap and market engagement initiatives. Enhancing AI research and development is another goal that Voxel51 hopes to achieve with this funding.</p>
 <p>Established in 2018, Voxel51 is an AI software company that develops tools for machine learning and computer vision researchers to quickly organize and experiment with their visual data collections.</p>
 
</div>

Voxel51 raises USD 30 million in Series B funding to accelerate product roadmap

PROMINENT_CATEGORY

<div>
 
 <p>ClearML, a comprehensive MLOps solution, has launched new features that provide AI orchestration and compute management support, compatible with platforms like Kubernetes, Slurm, PBS, and bare metal for efficient AI and ML task orchestration.</p>
 <p>The enhancements of ClearML automate numerous tedious tasks and offer greater control over AI infrastructure. This includes supporting several Kubernetes counterparts and bare metal, scheduling jobs, detailed compute resource allocation management, and open-source fractional GPU capabilities.</p>
 <p>These new features of ClearML are designed to handle the intricacies of AI deployments across different environments such as cloud, Edge, and on-premise data centers. The ultimate goal is to ease the management and control of AI infrastructure while enabling businesses to scale AI and machine-learning workflows effortlessly and efficiently.</p>
 
</div>

ClearML rolls out new AI orchestration and compute management capabilities

INCUMBENTS_PRESENT

<div>

<p>At the I/O 2024 conference, Google announced a new tool called "Firebase Genkit" for its Firebase platform. This open-source framework is intended to ease the process of creating AI-based applications, like content generation, summarization, text translation, and image generation.</p>

<p>Firebase Genkit is compatible with several programming languages, such as JavaScript and TypeScript, and will also include support for "Go" in the future. The tool lets developers infuse AI into their existing or new applications and offers a testing and deployment mechanism in Google’s serverless platforms.</p>

<p>Google designed Genkit with a focus on user-friendliness, extendability as per developers' requirements, and compatibility with an array of third-party open-source projects. Furthermore, Google's Project IDX will bolster the developer UI of Genkit, thereby improving its usability.</p>

</div>

Google announces 'Firebase Genkit' to simplify AI-powered app development at I/O 2024 conference

<div>
 
 <p>Google introduced its newest Tensor Processing Units (TPU), Trillium, at the annual Google I/O developer conference, which is set to be launched later in the year. Several improvements have been made to Trillium compared to its predecessors. These include an increase in compute performance per chip by 4.7 times, expanded matrix multiply units, heightened clock speed, and two times the memory bandwidth. The chips are also 67% more energy-efficient than the last generation.</p>
 <p>Unique to these chips is the inclusion of the third generation of SparseCore, an accelerator designed specifically for managing large-scale embeddings often seen in complex recommendation or ranking tasks. This allows Trillium TPUs to train models at a faster rate and serve them with reduced latency.</p>
 
</div>

Google unveils sixth generation of TPU chips at Google I/O conference

<div>
 
 <p>SensiML, a QuickLogic subsidiary focused on AI/ML software for IoT development, has launched an open-source AutoML solution for developing Edge AI/ML applications via its Analytics Studio application. The service adopts automated ML technology to allow developers with limited data science skills to create ML models for IoT microcontrollers and Edge system-on-a-chip (SoC).</p>
 <p>This strategy, according to the company, will foster the adoption of Edge AI/ML, improve end-user flexibility, and boost sales of SensiML's SaaS and its private-label tooling value. Founded in 2017, SensiML offers software for ultra-low-power IoT endpoints to transform raw sensor data into meaningful insights directly on the device. Its key product, the SensiML Analytics Toolkit, provides a complete platform for data collection, labeling, algorithm and firmware auto-generation, and testing.</p>

</div>

SensiML launches open-source AutoML solution to develop Edge AI/ML applications

<div>
 
 <p>AI development platform Snorkel AI has launched an upgrade for Snorkel Flow. This new update aims to simplify the process of coupling enterprise data with machine learning models for businesses.</p>
 <p>The latest improvement optimizes the integration of enterprise data. It provides connectivity to Vertex AI, Databricks Unity Catalog, and Microsoft Azure Machine Learning. These platforms allow for simplified data labeling, even facilitating coding-based labeling of multimodal data.</p>
 <p>Furthermore, the platform can be seamlessly linked with Language Learning Models (LLMs) such as Google’s Gemini 3 and Meta’s Llama 3. Such an integration offers businesses the flexibility to choose appropriate LLMs.</p>
 <p>Snorkel AI has also introduced a feature for role-based access controls (RBAC) specifically for Snorkel Flow account administrators. These controls allow administrators to manage the accessibility of enterprise data.</p>
 
</div>

Snorkel AI  unveils new data management capabilities

<div>
 <p>Nvidia has confirmed its plans to acquire Run:ai, a provider of Kubernetes-based workload management and orchestration software. Though the exact terms of the agreement were not revealed, the deal is reportedly valued at $700 million. Founded in Israel, Run:ai delivers a broad solution to enterprise customers to manage and optimize their compute infrastructure in on-premises, cloud, or hybrid set-ups by using an open platform built on Kubernetes.</p>
 <p>The platform integrates with different Kubernetes variants and numerous third-party AI tools and frameworks. Run:ai caters to some of the largest global enterprises from various industries by helping them efficiently manage their data center-scale GPU clusters through its platform.</p>
 <p>Upon completion of the acquisition, Nvidia plans to continue Run:ai's existing product line and integrate them into their DGX Cloud AI platform. This step is aimed at improving AI deployments, especially for GenAI implementations distributed over multiple data center locations, thereby enabling customers to utilize their AI computing resources more efficiently.</p>
 
</div>

NVIDIA acquires AI infrastructure management company Run:ai

<div>
 
 <p>Meta has announced the next version of its AI infrastructure, the Meta Training and Inference Accelerator (MTIA). This custom-made AI chip is designed to power both current and future Meta products.</p>
 <p>The new MTIA features enhanced compute power and memory bandwidth capabilities. It also supports the efficient architecture needed for GenAI applications, recommendation systems, and advanced AI research. The company is also developing several programs to expand MTIA's scope, specifically concerning support for GenAI workloads.</p>
 <p>The latest MTIA offers improved power and efficiency, boasting double the compute and memory bandwidth of its predecessor. These upgrades help to enhance Meta's platform content personalization abilities.</p>
 <p>In terms of hardware, the MTIA has also seen significant improvements. It features a rack-based design that can accommodate up to 72 accelerators, an elevated clocking speed, and an upgraded fabric between the accelerators and the host. The newest version also leverages PyTorch 2.0, integrating its performance benefits and related software features.</p>
 
</div>

Meta unveils next generation of Meta Training and Inference Accelerator (MTIA)

<div>
 
 <p>A data labeling and annotation platform, Sapien, has managed to gather a total of USD 5 million in a seed funding round. Notable participants of the funding round included Primitive Ventures, Artichoke Capital, and Yield Guild Games.</p>
 <p>The funds raised will primarily be used for the acceleration of product development.</p>
 <p>Sapien's main offering is a scalable annotation service for AI model training through its data labeling platform. The focus of the platform is the creation of a network for AI training data, which is generated by humans and provided to developers via an AI data marketplace. Sapien employs a novel approach to this system, involving gamification and rewards based on blockchain, to motivate its community of human labelers to come up with accurate annotations.</p>
 
</div>

Sapien raises USD 5 million in seed funding for product development

ANALYST_QUICK_TAKE

<div>

 <p>Intel introduced its most recent AI chip, Gaudi 3, which is tailored to improve the training and implementation of complex AI models such as those used by OpenAI's ChatGPT. The chip should be commercially accessible to clients in the third quarter, yet details regarding its price aren't disclosed yet.</p>
 <p>Gaudi 3 AI from Intel will power AI systems with a large number of accelerators, potentially in the tens of thousands. Intel asserts that its chip performs better than Nvidia's H100 GPUs.</p>
 <p>These new chips are Intel's attempt to meet specific enterprise-level Machine Learning (ML) and General AI (GenAI) requirements, as well as to tackle the scaling issues that enterprises regularly confront.</p>
 
</div> <p><b>Analyst QuickTake:<br>
</b> With the Gaudi 3 chip, Intel aims to disrupt and capture a significant portion of the rapidly expanding AI chip market, which is currently dominated by Nvidia, holding approximately 80% of the market share. Moreover, late last month, Intel and the US Department of Commerce <a href="https:<br>
//sp-edge.com/updates/27828"> signed a non-binding preliminary memorandum of terms </a> (PMT) for up to USD 8.5 billion in direct funding to Intel for commercial semiconductor projects under the CHIPS and Science Act.</p>

All Updates

Voxel51 raises USD 30 million in Series B funding to accelerate product roadmap

ClearML rolls out new AI orchestration and compute management capabilities

Google announces 'Firebase Genkit' to simplify AI-powered app development at I/O 2024 conference

Google unveils sixth generation of TPU chips at Google I/O conference

SensiML launches open-source AutoML solution to develop Edge AI/ML applications

Snorkel AI unveils new data management capabilities

NVIDIA acquires AI infrastructure management company Run:ai

Meta unveils next generation of Meta Training and Inference Accelerator (MTIA)

Sapien raises USD 5 million in seed funding for product development

Intel launches new AI chip Gaudi 3 to be commercially available from Q3

Machine Learning Infrastructure

Voxel51 raises USD 30 million in Series B funding to accelerate product roadmap

Contact us