AI Weekly Roundup: Quantum Advantage, Flying Cars, and Enterprise Supercomputing

NextSilicon’s Maverick-2 dataflow accelerator delivers up to 10× performance and ~60% lower power versus top GPUs, offering out-of-the-box AI/HPC acceleration without code rewrites across complex, real-world workloads. 
  

Tsinghua’s OFE² optical engine performs feature extraction at 12.5 GHz using light, enabling ultra-low-latency, low-power preprocessing that accelerates real-time AI tasks like imaging and trading.  
  

Google Quantum AI’s Quantum Echoes on Willow demonstrated verifiable quantum advantage (˜13,000× speedup), enabling precise molecular-structure computations that could rapidly unlock AI-driven discovery in chemistry and materials at scale. 
  

XPENG’s Aridge A868 flying car prototype entered trial production, integrating VLA 2.0 vision-to-action AI for embodied autonomy, with a mass-production assembly line capability signaling physical-AI industrialization.  
  

MIT Lincoln Laboratory unveiled TX-GAIN, a two-AI-exaflop supercomputer fusing 600+ GPUs, built to accelerate large-scale generative models and scientific simulations while cutting AI training power use by up to 80%. 

NASA and NVIDIA launched an AI-driven scientific-discovery platform using deep-learning and GPU-acceleration to process petabyte-scale datasets and accelerate breakthroughs in climate, astrophysics and Earth-system modelling. 
  

Google DeepMind launched Gemini 3 Pro, a natively multimodal model processing text, image and audio simultaneously, powering generative interfaces and deeper reasoning across modalities.  
  

Eli Lilly and NVIDIA launched the industry’s largest pharma-owned AI supercomputer, enabling model training at scale for drug discovery, manufacturing and clinical workflows across biopharma pipelines.  
  

Yale and Google DeepMind launched the Cell2Sentence-Scale 27 B AI model, capable of interpreting single-cell RNA data to generate biologically-validated hypotheses, accelerating precision medicine discovery in cancer and immunotherapy. 
  

NVIDIA, Microsoft, xAI and BlackRock launched a consortium agreement to acquire Aligned Data Centres for approximately $40 billion, scaling AI infrastructure access and locking future compute capacity for next-gen AI workloads.