Five ways data centers can save water
THE SO WHAT
Water use is joining power and latency as a first-class constraint for data center planning—especially for AI-heavy builds. If you're scaling inference or training, you need water and cooling efficiency in your site-selection and vendor scorecards, not just $/kWh.
READ THE SOURCE
MORE FROM THE WIRE
.png)
Deep & Emerging TechBritish Space Startup Launches Longevity Lab Into Orbit
Space-based biolabs feeding AI models on protein behavior tie longevity, AI, and space into a single R&D stack. If you’re in drug discovery or tools, assume your competitors will be training on data you literally cannot generate on Earth.
Deep & Emerging Tech'No single organisation can tackle it alone': VodafoneThree says its new security process blocked two million SMS fraud messages
VodafoneThree and Barclays cutting SMS fraud by 25% via shared intelligence shows telco–bank coalitions can actually move the fraud baseline. If you’re a comms or financial provider still running solo defenses, you’re effectively underwriting the attackers’ R&D.
Deep & Emerging TechSamsung’s Q2 profit jumps 19-fold to 89 trillion won on AI memory demand
A 19x profit jump to roughly 89.4 trillion won on AI memory demand underlines where the current value pool sits—in the components feeding model training and inference. If you’re upstream or downstream of HBM and memory, assume pricing and allocation risk will shape your roadmap more than model benchmarks this cycle.
Deep & Emerging TechExtreme SK Hynix Stock Swings Add Wild Card to $28 Billion Deal
SK Hynix’s volatility around a $28B US listing is a reminder that AI-chip exposure now brings hedge-fund-level swings to what used to be boring memory. Operators relying on these suppliers should diversify and lock in terms where possible—equity turbulence can spill into supply and pricing.