AI is too expensive

https://www.wheresyoured.at/ai-is-too-expensive/

#HackerNews #AI #Expensive #CostOfAI #AIChallenges #TechEconomics #FutureOfAI

Block eliminated over 4,000 jobs while posting record quarterly profits of $2.87B. CEO Dorsey attributed cuts to AI efficiency gains but provided no evidence of actual displacement. The stock jumped 24%, adding ~$8 billion in market value. That reward signal may incentivize other executives to follow the same playbook regardless of AI's actual impact on productivity.

#AIWashing #CorporateLayoffs #TechEconomics

https://www.implicator.ai/blocks-24-stock-jump-will-cost-more-jobs-than-its-ai-ever-could/

Anthropic's revenue jumped from $1B to $9-10B in 2025, with billions more added in January alone. CEO Amodei puts 50/50 odds on AI matching Nobel laureates within two years, but warns the industry faces trillion-dollar compute spending by 2028-29 with "bankruptcy risk if timing is off." The exponential may be nearing its end. #AI #Anthropic #TechEconomics

https://www.implicator.ai/anthropic-hit-10-billion-in-revenue-amodei-says-ai-nears-end-of-the-exponential/

MiniMax M2.5 scored within 0.6 points of Claude Opus 4.6 on key coding benchmarks while costing roughly 5% as much per token. Five Chinese AI labs shipped major models in a single week, suggesting the competition has shifted from capability gaps to cost optimization. Western labs face pressure to match prices without destroying research margins.

https://www.implicator.ai/minimax-matched-opus-at-5-of-the-price-the-ai-premium-is-dead/

#AI #CompetitiveDynamics #TechEconomics

Cloud pricing is not confusing by accident.
It is flexible, dynamic, usage-based, and perfectly aligned with uncertainty.

That uncertainty benefits providers more than users.
Predictability, stability, and clear limits still exist, but they cost extra.

You’re not paying for servers.
You’re paying for optionality.

#CloudComputing #SoftwareArchitecture #TechEconomics #EngineeringJudgment #ByernNotes

Investor Michael Burry and Anthropic co-founder Jack Clark debated AI economics in a rare written exchange. Key tension: Nvidia sold $400B in chips while AI app revenue sits below $100B. Anthropic admits it can't prove productivity gains from its own tools. METR study shows AI coding tools actually slow developers by 20%. The infrastructure-to-application gap raises questions about where value ultimately flows.

#AI #TechEconomics #Productivity

https://www.implicator.ai/the-escalator-problem-michael-burry-and-anthropics-co-founder-debate-whether-ai-is-a-bubble/

GPU prices are rising again in 2026—but this time, the cause isn’t crypto or short-term shortages. The real driver is memory. As VRAM costs surge and AI data centers absorb a growing share of global memory supply, GPUs are becoming structurally more expensive across both high-end and mid-range segments.

https://www.buysellram.com/blog/why-gpu-prices-are-rising-in-2026-how-memory-economics-and-ai-are-reshaping-the-graphics-market/

#GPUPrices #GPUMarket #AIHardware #VRAM #Nvidia #DataCenters #AIInfrastructure #MemoryMarket #GDDR7 #HBM #GPUs #HardwareLifecycle #TechEconomics #SupplyChain #tech

Global RAM Shortage Escalates: AI Training Consumes 70% New Capacity, Impacting Consumer Hardware

HBM/DDR5 prices double (Samsung +100%); GPU/laptop delays thru Q3'26. Geopolitics (TSMC/China) amplify risks

Supply chain: Stockpile/alt sources now.

Hardware buyers: Q1'26 window closing. Enterprise mitigation strategies?

https://buff.ly/1kzZNnN
#Semiconductors #SupplyChain #AIHardware #TechEconomics

Google's Gemini 3 Flash pricing reveals a shift in the AI industry's economics. While marketed as "a fraction of the cost" compared to Pro models, it costs 67% more than the Flash model it replaces - breaking four quarters of declining inference prices. The era of subsidized API access may be ending.

#AIpricing #LLMs #TechEconomics

https://www.implicator.ai/googles-gemini-3-flash-arrives-with-a-price-hike-disguised-as-a-discount/