AMD and KT, two industry giants, have spotlighted the future of AI model development by backing Moreh, an innovative AI software developer. Their support signals a significant shift in the AI landscape, emphasizing the importance of efficient and adaptable AI model creation.
Moreh’s vision is not just a pipe dream. With the recent investment, they’re poised to redefine how AI models are created and optimized. The backing from AMD and Korean telco KT underscores the potential of Moreh’s groundbreaking approach.
➜ Moreh’s Rise in the AI Software Arena
Moreh, a startup with roots in Santa Clara and Seoul, recently announced a significant milestone: securing $22 million in a Series B funding round. This boost, backed by giants like Advanced Micro Devices (AMD) and Korean telecommunications company KT, brings their total funding to a whopping $30 million.
At the heart of Moreh’s innovation is its flagship AI software, MoAI. Think of MoAI as a sibling to Nvidia’s CUDA. However, it is compatible with prevalent machine learning frameworks, such as Meta’s PyTorch and Google’s TensorFlow. This compatibility extends to applications and AI models previously exclusive to Nvidia.
➜ Addressing the AI Infrastructure Gap
Moreh isn’t just creating software for the sake of it. They’ve identified a gap in the market. While current AI software works wonders for small-scale AI models that use a handful of GPUs, it falls short for more extensive AI infrastructures. As AI matures and becomes a staple in enterprises, there’s a growing realization that many companies’ IT setups and data architectures aren’t cut out for training expansive AI models.
Gangwon Jo, Moreh’s co-founder and CEO, shed light on this in a chat with TechCrunch. He emphasized that Moreh’s solutions empower users, from AI developers to data center operators, to craft a more adaptable AI infrastructure. This adaptability is crucial, especially considering the global GPU shortage.
➜ Moreh’s Software: Bridging the Gap Between GPUs and AI Models
Diving deeper into the technicalities, Moreh’s software is a marvel. It allows GPUs and other AI chips, like neural processing units (NPUs), to run AI models, including behemoths like GPT-3 and T5, without altering code. For the uninitiated, while GPUs are the go-to for tasks like supercomputing and extensive data analysis, NPUs are tailor-made to speed up machine learning algorithms.
KT’s collaboration with Moreh since 2021 is a testament to the software’s prowess. They’ve been harnessing the power of AMD GPUs combined with MoAI software to create a scalable, cost-effective AI infrastructure. And the results? Stellar. KT found that Moreh’s technology works and excels, outperforming Nvidia’s DGX in speed and GPU memory capacity. Moreh’s tests showed that their platform, powered by AMD’s MI250 Instinct accelerator, had a 116% higher GPU throughput than Nvidia’s A100.
➜ Moreh’s Future Endeavors and Industry Impact
The startup isn’t resting on its laurels. After starting to generate revenue in 2021, they’ve set their sights on a revenue target of around $30 million by the end of 2023. And they have big plans for the recent funding. Moreh aims to channel the funds into research, product expansion, and team growth. They currently boast a team of 70, but with the backing of investors like South Korean VC firms Smilegate Investment and Forest Partners, they’re poised for even more growth.
“The AI software ecosystem supporting AMD AI hardware continues to grow, providing choice for data scientists and other users of AI as they build the AI models and solutions that will drive the continued growth of this industry,”
Brad McCredie from AMD weighed in on Moreh’s journey.
Moreh’s journey is a testament to the evolving landscape of AI software development. Their focus on creating adaptable, efficient tools for AI model development is innovative and necessary in today’s tech-driven world. As they continue to push boundaries, one can’t help but be reminded of the importance of technological adaptability. For those keen on diving deeper into AI advancements, NeuralWit offers many insights.