Hi Synapticians!

Today’s edition is all about AI reshaping everything from health to self-driving cars, with a hefty dose of compute power drama in the mix.

Let’s start with Gleamer, the French AI startup determined to supercharge medical imaging. Instead of reinventing the wheel, they’re acquiring existing AI players like Pixyl and Caerus Medical to fast-track their expansion into MRI diagnostics. With millions of exams already under their belt, their vision of AI-driven preventive imaging sounds promising—if they can tackle the global shortage of radiologists. Efficiency is great, but who’s going to interpret all these AI insights?

Meanwhile, Hugging Face and Yaak are going full throttle with self-driving tech, launching L2D, the largest open-source dataset for autonomous driving. Sixty EVs roaming German streets for three years have produced a goldmine of data on real-world driving behavior. This open-source approach could speed up AI’s road skills, but will it be enough to nudge driverless cars into mainstream reality?

Then there’s OpenAI, which has just secured a jaw-dropping $12 billion worth of computing power from CoreWeave. Clearly, OpenAI believes that more compute equals better AI, but not everyone is sold—Microsoft’s CEO has openly warned that hardware alone won’t guarantee success. Throw in CoreWeave’s recent acquisition of Weights & Biases, and we’ve got ourselves a full-blown power play in the AI infrastructure race.

Finally, Dassault Systèmes is using generative AI to make virtual hearts more personal than ever. Their Living Heart project now integrates AI to improve treatment predictions and reduce reliance on human and animal testing. With FDA support, these ""in silico"" trials are gaining credibility. If they extend this tech to other organs, AI could completely redefine medical research.

From supercharged medical imaging to self-learning cars and relentless AI infrastructure wars, today’s stories prove that AI isn’t just evolving—it’s accelerating fast. Ready to dive in?

Top AI news

1. OpenAI secures $12B in compute power with CoreWeave
OpenAI has secured nearly $12 billion in computing power from CoreWeave, reinforcing its belief that more compute leads to better AI. The deal includes a $350 million investment, making OpenAI a CoreWeave shareholder. This move aligns with OpenAI’s strategy of scaling AI through massive infrastructure, though Microsoft’s CEO warns that compute alone doesn’t guarantee success. Additionally, CoreWeave’s acquisition of Weights & Biases strengthens its AI development ecosystem. The deal highlights the ongoing race for AI infrastructure dominance and raises questions about the long-term sustainability of compute-heavy AI strategies.

2. Dassault Systèmes enhances virtual heart models with AI
Dassault Systèmes has integrated generative AI into its "Living Heart" project, creating highly personalized virtual heart models. These AI-powered simulations enhance treatment predictions, reduce reliance on human and animal trials, and accelerate medical device development. Supported by the FDA’s "ENRICHMENT Playbook," this innovation strengthens the credibility of in silico clinical trials. The technology could extend to other organs, revolutionizing precision medicine and regulatory processes while reducing costs and time-to-market for medical innovations.

3. Gleamer expands into MRI with AI acquisitions
Gleamer, a French AI startup specializing in medical imaging, is expanding into MRI by acquiring Pixyl and Caerus Medical. Instead of building from scratch, it integrates existing AI technologies to accelerate progress. With 2,000 institutions in 45 countries using its software and 35 million exams processed, Gleamer aims to improve diagnostic accuracy. The company envisions a future where AI-driven preventive imaging becomes standard, but the shortage of radiologists remains a challenge. By automating and orchestrating diagnostics, Gleamer seeks to address this growing demand.

Bonus. L2D: The largest open-source dataset for autonomous driving AI
Hugging Face and Yaak have launched L2D, the world’s largest open-source dataset for autonomous driving AI. Collected over three years with 60 electric vehicles in 30 German cities, L2D provides multimodal data on expert and student driving behaviors. It includes natural language instructions, sensor data, and real-world driving scenarios, making it a valuable resource for training AI models. Integrated with Hugging Face’s LeRobot pipeline, L2D allows the AI community to explore, contribute, and refine driving episodes. This initiative aims to enhance spatial intelligence and accelerate the adoption of self-driving technology.

Tweet of the Day

Theme of the Week

DeepSearch - The Startup
​Perplexity AI's recent introduction of "Deep Search" enhances traditional search by offering comprehensive, context-aware insights drawn from diverse, authoritative sources. This advanced feature leverages powerful language models to provide detailed, accurate, and conversational answers, significantly improving how users engage with complex queries.

Stay Connected

Feel free to contact us with any feedback or suggestions—we’d love to hear from you !