Our recent visit to NVIDIA GTC was an enlightening experience. We were thrilled to honor them as the company of the year. Their journey is a testament to the power of relentless dedication, which has led to the unveiling of opportunities that even the founders hadn't initially envisioned. Silicon Valley is a hub of such success stories, and it's always intriguing to delve into what makes these companies thrive. As an author, I find it enriching to share the wisdom I've gained from investors. Companies that enhance the efficiency of the development process have a promising future. Giants like Apple, Google, Microsoft, and Amazon have set a high bar in this regard, not only empowering developers but also fostering a professional community through top-notch training.

illustration

Copyright© Schmied Enterprises LLC, 2024.

Established in 2021, Smartcow is a multinational corporation with leadership hailing from Malta, Taiwan, Dubai, and India. The company specializes in software-defined hardware solutions, with a particular focus on Jetson-based edge AI solutions for image processing.

Avermedia is a seasoned player in the field of capture devices and PC cameras, boasting a rich history and extensive experience.

Optra's edge devices, with their aesthetic appeal and GPU performance ranging from 0.5 to 20 TFLOPS, are designed to be a stylish addition to any workspace.

Lanner Electronics offers a comprehensive product portfolio that includes rack mount, edge, and extension devices. Their hardware solutions cater to a wide range of needs, from customer tracking on shop floors to the transfer of over 100 gigabits of video.

Numurus offers a variety of Jetson-based edge devices, some of which are even capable of functioning underwater.

Liqid and Wiwynn provide datacenter solutions, with Liqid offering up to 20 GPUs per rack and Wiwynn focusing on AI and graphics applications featuring NVidia GPUs.

Boston, a UK-based company, offers a wide range of datacenter solutions, including products capable of running H100s. Centersquare, a global data center company, offers storage solutions for AI training. Bulk, on the other hand, specializes in building and maintaining network infrastructure and datacenters in Scandinavia.

Wolf Advanced Technology and Colfax International offer datacenter products for graphics applications, with Colfax bringing 36 years of experience to the table. Future Tech Enterprise provides solutions for AI and graphics applications, specializing in serving government vendors.

Hyve focuses on motherboard design, the heart of every server, while Axiomtek offers industrial motherboards and accessories for monitoring, manufacturing, and automation.

Concurrent Real-time offers a variety of real-time Linux solutions for embedded projects, including an FPGA workbench solution for high-end development needs. Forecr and Siliconhighway offer products and solutions based on the NVIDIA Jetson architecture, catering to AI and graphics edge applications.

A number of companies have been inspired by this new class of devices, with graphics being a primary focus. Hololight offers streaming solutions for AR, VR, and CAD, enabling rendering in the cloud. Reve computer provides server and workstation solutions for immersive applications, such as their flagship product, Renderbeast. Rendered.ai offers generated image data for AI applications, providing a service that includes hardware streaming generated content starting at $5000 a month. Leiainc offers a patented display solution that can transform a 2D display into a 3D scene for CAD and similar applications.

Streaming services are another significant group of vendors. Streaming is a complex task, not due to graphics handling, but rather the large bandwidth requirement traffic that is still challenging to manage. This is particularly true for content that requires generation by AI or simulation, which is inherently interactive and requires super low latencies and very small jitter of 10-20ms. This is currently only feasible when the user is located close to an edge datacenter like Lumen. Notable providers in the graphics streaming and GPU cloud space today include Pureweb, Crusoe, Iren, Paperspace, Foundries.io, and Innoactive.

Hosting GPUs can be costly, making renting them from a cloud provider a more economical option. As a result, the GPU inference industry is likely to shift towards serverless solutions. Exhibitors in this space included Cloudflare, Covalent, and Ubiops.

The exhibition also featured many companies from the "fun tech" sector. For robotics applications, arrow.com is a recommended resource. Cogniteam offers a cloud robotics platform, and Digital Realty provides a global data center platform that offers colocation, connectivity, and AI solutions. Trifork is a company to consider for implementing ideas, while secedge.com provides traditional security and update services for edge devices.

The author concludes that the GPU industry has always been at the forefront of representation and artistic value, attracting equity financing and sponsorship. Many companies have leveraged this edge to create unique value products. The advent of AI has transformed the landscape, positioning NVIDIA as a core infrastructure supplier connected to basic public industries. This transition to widespread acceptance of such solutions in daily life will be a slow and challenging process, not due to technology, but societal absorption rates of new innovations.

AI avatars, such as videos narrated generated from text, are a particular interest of the author. The South Korean company klleon.io, which offers such a service, is highly recommended.

In terms of platforms, Kubernetes is currently the accepted choice, but competitors exist, with run.ai being a notable presence at the conference.

Exhibiting AI solution developers included Slalom, Xiilab, Writer, DECI AI, Upstage, Deepset, Edgeless Systems, Latent AI, Marvik, Kinetia, and Airjet. Differentiation is challenging in this space, but one thing is clear: NVIDIA is here to stay. The evolution of these companies in the future will be intriguing to observe.