AI inference

Chinese firms pay ~$1M per Nvidia B300 server—double U.S. prices—amid export restrictions, while China's AI token usage surges to 32% globally.

$NVDA regains $5 trillion market cap with potential to become world's first $10 trillion company within three years, driven by AI inference dominance.

Nebius acquires Eigen AI for $643M to strengthen AI inference capabilities. Stock rises 2.97% in premarket trading despite elevated 20.42% short interest.

Rambus stock plummets 10.48% after Q1 EPS miss and weak Q2 guidance, despite beating revenue expectations.

$MU gains 4.5% as Intel's earnings beat signals strong AI inference demand, bolstering prospects for high-bandwidth memory chips.

Digital Realty commits S$7 billion to Singapore data center expansion and AI innovation, positioning the city as Asia Pacific's infrastructure hub.

Nvidia and Broadcom lead AI chip market as focus shifts from training to inference. Nvidia targets $1T sales by 2027; Broadcom forecasts $100B+ AI revenue. Similar valuations, different growth profiles.

Google's TurboQuant compression tech cuts DRAM needs 6x and boosts speeds 8x, rattling memory stocks like $MU despite likely long-term demand benefits.

Micron stock fell 18% post-earnings despite 196% revenue and 682% EPS growth, driven by pricing and AI compression concerns. Supply tightness expected through 2026.

Intel launches Arc Pro B70 GPU at $949, undercutting Nvidia and AMD alternatives by hundreds of dollars to capture growing local AI workload market.

Nvidia's $1T chip order pipeline through 2027 hints at massive AI inference opportunity, but muted stock reaction reflects already-high expectations and valuation pressures.

Broadcom's custom AI accelerators grow 140% annually, targeting $100B+ revenue by 2027. Trading at 30x forward earnings, the premium valuation reflects faster growth than Nvidia in the expanding AI chip market.

Nvidia restarts H200 chip production for China following regulatory approval, potentially unlocking a $50 billion market and boosting growth.

Nvidia forecasts $1 trillion revenue by 2027 via AI inference, but investor skepticism persists amid intensifying competition and custom chip threats.

Penguin Solutions launches OriginAI Factory Platform integrating MemoryAI and ICE ClusterWare with NVIDIA GPUs to optimize enterprise AI inference, targeting financial services, healthcare, and retail sectors.

Penguin Solutions debuts MemoryAI, an 11TB CXL-based KV cache server offering 10x faster AI inference speeds than NVMe, compatible with NVIDIA's architecture.

$VCIG launches AI Compute Treasury strategy to accumulate NVIDIA GPU infrastructure, targeting growing AI inference demand in market projected to reach $394.5B by 2030.

VCIG Global launches AI Compute Treasury strategy to accumulate NVIDIA GPU infrastructure, supporting enterprise AI inference workloads through scalable revenue reinvestment model.

CoreWeave's Perplexity deal validates shift from AI training to inference workloads. $66.8B backlog and $30-35B 2026 capex support analyst targets of $124.34.

AI inference spending surges toward $700B by 2026, challenging Nvidia's dominance as cloud providers develop custom chips and competitors like Broadcom and AMD gain ground.