I love adaptive reuse. At 150 Hayes, they are turning an ugly vacant office building to a beautiful 105-unit residential building (1,2,3 BR's) for ~$350/sf (construction cost) + $225/sf (acquisition cost). The project only pencils because SF waived affordable-housing requirements for office-to-resi conversions. $575/sf (construction + acquisition cost) seems like a pretty good deal for what looks like a nice building. The same guys did a great job on 100 Van Ness.

.@elonmusk on why full rocket reuse matters. Not recovering the upper stage "is like throwing away a medium-sized jet on every flight." "If you had to throw away the planes after every flight, it would be far too expensive, and air flight would not be a realistic option." 🚀 🎥 CGTN / CMG interview

BREAKING: SpaceX has set a new reuse record! A Falcon 9 fairing half has flown for the 40th time, the first ever in SpaceX history. The nose-cone shell, once discarded in the ocean after a single launch, has now flown 40 times. Incredible.

SpaceX just broke its own booster reuse record again B1067 completed its 37th launch and 37th landing....surpassing the previous record of 36 flights, which it also held And it happened on SpaceX’s 100th Falcon 9 launch of 2026

BREAKING: SpaceX just smashed a new rocket-reuse record. Today, Falcon 9 booster B1067 became the first booster ever to complete 37 launches and landings. • Launched for a record 37th time • Landed successfully for a record 37th time

💾 TL;DR: reuse a KV cache computed on another node's GPU straight through CXL memory, up to 36.6x faster. Seagate proposes a Kubernetes-native way to share CXL memory across the cluster. Title: Composable CXL Memory as a Kubernetes-Native Shared Memory for LLM Serving URL: 📌 Highlights 🧩 A Kubernetes DRA driver turns CXL regions into schedulable cluster resources 🗂 No Redis/etcd needed, KV slots are tracked inside the region itself 🚀 Cross-node reuse hits 36.6x speedup over recompute at 32K-token prefixes 🎯 Hit rates reach 95.4-99.5% ⚖️ Cross-node vs same-node latency differs by only 1-4% 🐢 Still 1.15-1.70x slower than an on-GPU VRAM prefix cache hit 🔧 Measured 5.8 GB/s out of a 27 GB/s fabric ceiling, room to grow with zero-copy DMA For serving that needs to reuse long prefixes across GPU nodes, this feels like a genuinely practical option. #LLMInference# #Kubernetes#

China is accelerating the establishment of rules governing the recycling and reuse of wind and solar equipment and aims to set up a closed-loop system for waste from the industry before 2030, China Central Television reported on Sunday