my technical predictions have been largely correct for 25 years and i believe understand the architectures well, so feel comfortable making predictions.
COMPUTING
-
AI Agent Syncs 2.5K Files to Cloud Drive Storage
By
–
5.5 in codex just synced all my memories and files to a cloud agent drive 2.5k files
-
Inference Costs Approaching Electricity Parity as AI Scales
By
–
they got no clue, and quote from NVIDIA taken out of context. classic media. they are as much domain experts as snake oil sales men. We’re running, no correction, we’re sprinting to inference being at cost with electricity production. inference will be as abundant as air
-
AI Inference Costs Dropping 5-10x Yearly, Challenging Compute Scarcity Narratives
By
–
When I read articles like this one in Forbes, I wonder if they realize that AI inference is currently becoming about 5-10x cheaper each year, with occasional jumps of 10-100x for certain capabilities. It may be true of this year and maybe the year after that "compute is far
-
Niantic’s 3D World Model Enables Precise Camera-Based Localization
By
–
Niantic has the 3D model of the entire world. I got a demo a few months ago, it has a digital twin of the entire world down to about a millimeter per voxel.
— Robert Scoble (@Scobleizer) 29 avril 2026
As soon as a camera turns on that uses it it will know where you are and what you are looking at.
The only variable is… https://t.co/qma3sWg2YFNiantic has the 3D model of the entire world. I got a demo a few months ago, it has a digital twin of the entire world down to about a millimeter per voxel. As soon as a camera turns on that uses it it will know where you are and what you are looking at. The only variable is
-
DeepSeek v4 Demonstrates SOTA Long Context Efficiency Without Benchmarking
By
–
IMO DeepSeek v4 demonstrated utter confidence and competence by not benchmaxxing, not focusing on some BS final run cost, not even spending inference-optimal compute. just showed up, demonstrated SOTA long context efficiency techniques (CSA, HCA, mHC, flash at 8% cost of pro,
-
Claude Opus 4.7 dominates reasoning and code benchmarks
By
–
Deep reasoning and code precision: Claude Opus 4.7 → SWE-Bench Pro: 64.3% → GPT-5.5: 58.6% → MCP Atlas: 79.1% → GPT-5.5: 75.3% → GPQA Diamond, HLE (with and without tools), FinanceAgent v1.1: all Opus 4.7 When the task requires architectural thinking
-
Iterative Material Design Process and Composition Analysis
By
–
Why not more frequently? The process for designing new materials can take many iterative learning cycles, where materials’ compositions are analysed, and resulting effects are closely scrutinised.
-
Revolutionary Materials Transform Semiconductor Industry Once Generation
By
–
Introducing new materials into industrial processes can be a revolution for customers across industries. In the semiconductor space for example, a new, gamechanging material is introduced about once in a generation… pic.twitter.com/wgKZN1tLZ1
— Max Welling (@wellingmax) 29 avril 2026Introducing new materials into industrial processes can be a revolution for customers across industries. In the semiconductor space for example, a new, gamechanging material is introduced about once in a generation…
-
Data Centers Optimize Energy for Growing AI Demand
By
–
Data centers hinge on precise energy orchestration across cooling and power. As AI demand strains capacity, design choices hit cost and uptime, pushing efficiency into daily operations. Source @EyeOnTech_TT via @antgrasso