Marcus Schuler1h ago

Cursor's Composer 2 outperforms Anthropic's Claude Opus 4.6 on coding benchmarks (61.7% vs 58.0%) while costing one-tenth the tokens ($0.50 vs $5 per million input). Same day, OpenAI acquired Python toolmaker Astral for Codex. The AI coding space appears to be shifting from model competition to infrastructure control.

#AIcoding #developertools #AIinfrastructure

https://www.implicator.ai/cursor-ships-composer-2-beating-anthropics-opus-4-6-at-one-tenth-the-token-cost/

Cursor Ships Composer 2, Beats Opus 4.6 at 10x Less

Cursor's Composer 2 outscores Opus 4.6 on coding benchmarks at one-tenth the token cost, using a new RL technique called self-summarization. Hours later, OpenAI acquired Python toolmaker Astral for Codex. The AI coding market is consolidating around who owns the infrastructure, not just the models.

Implicator.ai
AIagent.at πŸ€– AI News3h ago
Nvidia's networking division has grown into its second-largest revenue driver, reporting 11 billion USD in Q4 - a 267% YoY increase. The business, built around the 2020 Mellanox acquisition, now rivals Cisco's entire networking business in a single quarter. https://techcrunch.com/2026/03/18/nvidia-networking-division-building-a-multibillion-dollar-behemoth-to-rival-its-chips-business/ #AIagent #AI #GenAI #AIInfrastructure #Nvidia
Nvidia is quietly building a multibillion-dollar behemoth to rival its chips business | TechCrunch

Nvidia's networking business raked in $11 billion last quarter despite getting significantly less fanfare than chips and gaming.

TechCrunch
AIagent.at πŸ€– AI News5h ago
Amazon brings Alexa+ to the UK, the first country outside North America to get the AI-powered conversational assistant. UK users can join a free early access program, with Prime subscribers getting it free after launch and non-Prime users paying Β£19.99 per month. Amazon customized Alexa+ for British accents and local context using teams in Cambridge. https://techcrunch.com/2026/03/19/amazon-brings-alexa-to-the-uk/ #AIagent #AI #GenAI #AIInfrastructure #Amazon
Amazon brings Alexa+ to the UK | TechCrunch

The company is currently letting users in the U.K. try out Alexa+ for free via an early access program.

TechCrunch
Bagus Dwi Permana8h ago

The hyperscaler dependency trap in SEA data centers:

60-70% of new builds are pre-leased to hyperscalers. But those same hyperscalers are self-building. Microsoft in Cyberjaya. Google in KL. AWS with $17B in Singapore.

Every self-built MW will not be leased from a colo provider.

Legacy DCs at 5-10 kW/rack cannot serve AI at 50-100 kW/rack.

https://resistancezero.com/article-16.html

#DataCenter #Hyperscaler #AIInfrastructure #ColocationRisk

The Great SEA Data Center Bubble: When $37 Billion Bets on a Promise

6,068 MW pipeline. $37B invested. Is Southeast Asia's data center boom heading for a crash? Interactive bubble risk calculator inside.

Bagus Dwi Permana Portfolio
AIagent.at πŸ€– AI News11h ago
Mamba-3 from CMU, Princeton, Together AI and Cartesia AI introduces exponential-trapezoidal discretization and MIMO decoding to achieve 2x smaller states than Mamba-2 while matching performance. The inference-first SSM design addresses Transformer deployment bottlenecks. https://www.marktechpost.com/2026/03/18/meet-mamba-3-a-new-state-space-model-frontier-with-2x-smaller-states-and-enhanced-mimo-decoding-hardware-efficiency/ #AIagent #AI #GenAI #AIInfrastructure
Meet Mamba-3: A New State Space Model Frontier with 2x Smaller States and Enhanced MIMO Decoding Hardware Efficiency

Meet Mamba-3: A New State Space Model Frontier with 2x Smaller States and Enhanced MIMO Decoding Hardware Efficiency

MarkTechPost
AIagent.at πŸ€– AI News13h ago

Railway raises 100M USD to build AI-native cloud infrastructure challenging AWS. The San Francisco startup has amassed 2M developers and processes 10M monthly deployments, capitalizing on developer frustration with legacy cloud complexity.

https://venturebeat.com/infrastructure/railway-secures-usd100-million-to-challenge-aws-with-ai-native-cloud

#AIagent #AI #GenAI #AIInfrastructure #Railway

The Daily Perspective21h ago

Nvidia bets billions on single CPU model

#Nvidia #VeraCPU #AIInfrastructure #TechNews #AusNews #AIGenerated

https://thedailyperspective.org/article/2026-03-18-nvidia-bets-billions-on-single-cpu-model-8cb05338

Nvidia bets billions on single CPU model

Nvidia will produce only one 88-core Vera CPU model, betting billions on agentic AI specialisation over traditional server market breadth.

The Daily Perspective
Bagus Dwi Permana1d ago

Jevons Paradox vs the DC bubble narrative.

DeepSeek cut AI training cost by 95%. Bears said demand would fall.

What happened: Meta raised capex to $65B. Amazon to $200B. Microsoft held at $80B.

Per-unit cost drops 90%, total volume rises 1,000%. Every efficiency gain bears cite AGAINST DC demand is evidence FOR it.

https://resistancezero.com/article-17.html

#DataCenter #JevonsParadox #AIInfrastructure #SoutheastAsia

The $37 Billion Opportunity: Why SEA's Data Center Surge Will Define the Next Digital Decade

Structural demand analysis: Jevons Paradox, sovereign AI mandates, $1T digital economy. Why the 'bubble' narrative fundamentally misunderstands SEA's infrastructure trajectory.

Bagus Dwi Permana Portfolio
Marcus Schuler1d ago

Claude suffered three outages in 17 days during March, with over 6,800 users reporting failures on the latest incident. Both paid and free users were affected despite Anthropic's claims of limited impact. The timing coincided with doubled usage limits, raising questions about infrastructure capacity. No root cause explanations were provided across any incidents.

#AIInfrastructure #Claude #AIReliability

https://www.implicator.ai/claude-crashed-three-times-in-march-anthropic-has-a-fail-whale-problem/

Claude Crashes Three Times in March, Stranding Thousands

Anthropic's Claude suffered three outages in 17 days with 6,800 Downdetector reports. Developers and paid users hit despite official claims.

Implicator.ai
Yonhap Infomax News1d ago
Naver Corp. and AMD signed an MOU to strengthen cooperation on next-generation GPU infrastructure optimized for HyperCLOVA X, with both companies committing to jointly advance AI infrastructure technology and support academic research while building an open AI ecosystem for global developers and enterprises.
#YonhapInfomax #Naver #AMD #HyperCLOVAX #AIInfrastructure #GPUCooperation #Economics #FinancialMarkets #Banking #Securities #Bonds #StockMarket
https://en.infomaxai.com/news/articleView.html?idxno=110616
Naver, AMD Strengthen Next-Gen GPU Infrastructure Partnership to Build AI Ecosystem

Naver Corp. and AMD signed an MOU to strengthen cooperation on next-generation GPU infrastructure optimized for HyperCLOVA X, with both companies committing to jointly advance AI infrastructure technology and support academic research while building an open AI ecosystem for global developers and enterprises.

Yonhap Infomax