FlexAI frames itself as a platform-as-a-service for AI teams that need to train, fine-tune, and serve models without turning every ML sprint into MLOps firefighting. The idea is workload right-sizing: pick the smallest viable cluster shape for each job, reduce idle GPU time, and keep iteration loops tight, with a stated goal of cutting typical costs to around 30% while improving time-to-train for real projects. https://www.flex.ai/
—
HDMI® Technology is the foundation for the worldwide ecosystem of HDMI-connected devices; integrated with displays, set-top boxes, laptops, audio video receivers and other product types. Because of this global usage, manufacturers, resellers, integrators and consumers must be assured that their HDMI® products work seamlessly together and deliver the best possible performance by sourcing products from licensed HDMI Adopters or authorized resellers. For HDMI Cables, consumers can look for the official HDMI® Cable Certification Labels on packaging. Innovation continues with the latest HDMI 2.2 Specification that supports higher 96Gbps bandwidth and next-gen HDMI Fixed Rate Link technology to provide optimal audio and video for a wide range of device applications. Higher resolutions and refresh rates are supported, including up to 12K@120 and 16K@60. Additionally, more high-quality options are supported, including uncompressed full chroma formats such as 8K@60/4:4:4 and 4K@240/4:4:4 at 10-bit and 12-bit color.
—
On the training path, you connect a GitHub repo, point to your requirements file and entry point, then choose node count and accelerator count. Instead of manually aligning drivers, CUDA, PyTorch builds, containers, and dependency pinning, FlexAI automates the environment so scaling from 8 GPUs to 16 is a config edit rather than a re-install cycle. The platform is described as deployable across regions (including France and the US) and able to run on AWS, GCP, or Azure when you want it inside your broader stack.
Inference gets treated like a sizing and orchestration problem rather than “pick a GPU and hope”: an Inference Sizer asks for throughput targets (requests per second), token sizes, and model class, then recommends GPU SKU and GPU count based on benchmarking. The demo highlights fractional GPUs (down to slices such as 1/7), autoscaling bounds, and an OpenAI-compatible API endpoint you can drop into an app, plus built-in observability for latency, throughput, and utilization as a single metric.
A practical driver here is the move away from frontier-model APIs once cost curves and hallucination risk become product liability: teams start with hosted endpoints, then migrate to fine-tuned open models (or train from scratch for narrow domains like space, legal, or healthcare) where behavior, evaluation, and data control matter. Filmed at Web Summit Lisbon 2025, the interview also sketches FlexAI’s company arc: a public $30M seed round (Alpha Intelligence Capital, Elaia, Heartcore), a roughly 30-person team spread across France, India, and the US, and plans to build more community presence via Station F in Paris in January.
The longer-term bet is heterogeneous compute without lock-in: support for multiple GPU families (NVIDIA H100/H200 and newer Blackwell-era options like GB200/B200, plus AMD paths such as MI300-class inference), and the ability to route workloads across clouds while keeping the developer surface area stable. Combine that with utilization-driven scheduling and you can imagine carbon-aware placement — steering big training runs to cheaper, lower-carbon grids when deadlines allow — without forcing a rewrite of pipelines, which is the strategic path.
I’m publishing about 90+ videos from Embedded World North America 2025, I upload about 4 videos per day at 5AM/11AM/5PM/11PM CET/EST. Join https://www.youtube.com/charbax/join for Early Access to all 90 videos (once they’re all queued in next few days) Check out all my Embedded World North America videos in my Embedded World playlist here: https://www.youtube.com/playlist?list=PL7xXqJFxvYvjgUpdNMBkGzEWU6YVxR8Ga
This video was filmed using the DJI Pocket 3 ($669 at https://amzn.to/4aMpKIC using the dual wireless DJI Mic 2 microphones with the DJI lapel microphone https://amzn.to/3XIj3l8 ), watch all my DJI Pocket 3 videos here https://www.youtube.com/playlist?list=PL7xXqJFxvYvhDlWIAxm_pR9dp7ArSkhKK
Click the “Super Thanks” button below the video to send a highlighted comment under the video! Brands I film are welcome to support my work in this way 😁
Check out my video with Daylight Computer about their revolutionary Sunlight Readable Transflective LCD Display for Healthy Learning: https://www.youtube.com/watch?v=U98RuxkFDYY



