A field guide · Updated June 2026

The most consequential technology in human history is being built right now.

We have entered the **Silicon Epoch**. In the next decade, artificial intelligence will rewrite science, medicine, work, geopolitics, and the texture of daily life. This is a living, deeply researched guide to *who* is building it, *how* it works, the *grid* that powers it, and *where* it is going.

GPT-5.4 (June 2026) · Claude Fable 5 (June 9, 2026) · Claude Mythos 5 (June 2026) · Gemini 3.1 Pro (June 2026) · DeepSeek V4 MoE ~1T params · Meta Muse Spark Pivot · Qwen 3.7-Max (May 2026) · Kimi K2.5 Open Weights · TSMC 2nm N2 Volume Production (Q4 2025) · Microsoft Crane Clean Energy Center 835 MW (2027) · Google Kairos Power SMR Deal (500 MW) · DeepSeek-R1 Test-Time Scaling Shift · SWE-bench Pro Opus 4.8 69.2% · Claude Code $2.5B ARR · Factorio baseline 29.1 (Claude 3.7 Sonnet) · GPT-5.4 (June 2026) · Claude Fable 5 (June 9, 2026) · Claude Mythos 5 (June 2026) · Gemini 3.1 Pro (June 2026) · DeepSeek V4 MoE ~1T params · Meta Muse Spark Pivot · Qwen 3.7-Max (May 2026) · Kimi K2.5 Open Weights · TSMC 2nm N2 Volume Production (Q4 2025) · Microsoft Crane Clean Energy Center 835 MW (2027) · Google Kairos Power SMR Deal (500 MW) · DeepSeek-R1 Test-Time Scaling Shift · SWE-bench Pro Opus 4.8 69.2% · Claude Code $2.5B ARR · Factorio baseline 29.1 (Claude 3.7 Sonnet) ·
GPT-5.4 (June 2026) · Claude Fable 5 (June 9, 2026) · Claude Mythos 5 (June 2026) · Gemini 3.1 Pro (June 2026) · DeepSeek V4 MoE ~1T params · Meta Muse Spark Pivot · Qwen 3.7-Max (May 2026) · Kimi K2.5 Open Weights · TSMC 2nm N2 Volume Production (Q4 2025) · Microsoft Crane Clean Energy Center 835 MW (2027) · Google Kairos Power SMR Deal (500 MW) · DeepSeek-R1 Test-Time Scaling Shift · SWE-bench Pro Opus 4.8 69.2% · Claude Code $2.5B ARR · Factorio baseline 29.1 (Claude 3.7 Sonnet) · GPT-5.4 (June 2026) · Claude Fable 5 (June 9, 2026) · Claude Mythos 5 (June 2026) · Gemini 3.1 Pro (June 2026) · DeepSeek V4 MoE ~1T params · Meta Muse Spark Pivot · Qwen 3.7-Max (May 2026) · Kimi K2.5 Open Weights · TSMC 2nm N2 Volume Production (Q4 2025) · Microsoft Crane Clean Energy Center 835 MW (2027) · Google Kairos Power SMR Deal (500 MW) · DeepSeek-R1 Test-Time Scaling Shift · SWE-bench Pro Opus 4.8 69.2% · Claude Code $2.5B ARR · Factorio baseline 29.1 (Claude 3.7 Sonnet) ·
78%

Of organisations regularly use generative AI, up from 55% in 2023 (McKinsey 2025)

140T

Daily token volume call rate in China as of Q1 2026 — a 1,400× jump in two years

69.2%

Top score on SWE-bench Pro (Claude Opus 4.8, June 2026) — up from ~10% two years ago

$2.5B

Claude Code's annualized revenue in February 2026 — just 9 months post-release

Frontier AI Labs & Ecosystem covered in this guide
OpenAIOpenAI
AnthropicAnthropic
DeepMindDeepMind
GrokxAI
MetaMeta MSL
DeepSeekDeepSeek
MistralMistral AI
CohereCohere
MoonshotAIMoonshot
QwenAlibaba Qwen
ByteDanceByteDance
BaiduBaidu Ernie
ZhipuZhipu GLM
TencentTencent
Field Guide Chapters

Thirteen vectors of the intelligence explosion.

01 · Compute Core

Silicon fundamentals & packaging

TSMC 2nm GAA nanosheets, Intel 18A, backside power, and the critical CoWoS advanced packaging capacity bottlenecks powering high-performance accelerators.

Inspect the silicon
02 · The Mechanics

How AI actually works

Transformers, self-attention, GQA, Dense vs Mixture of Experts (MoE), and linear-complexity alternatives like Mamba 3 and Linear Recurrent Units.

Open the engine bay
03 · Frontier Labs

Every major lab & flagship model

From OpenAI's GPT-5.4 family and Google's Gemini 3.1 to Anthropic's Claude Fable 5, Meta's Maverick, DeepSeek V4, Qwen 3.7-Max, and Zhipu GLM-5.

Browse the labs
04 · Use Cases

Reasoning, science, and coding

SWE-bench Pro coding performance, AlphaGeometry 2 IMO gold-medalist math, AlphaFold 3 biomolecular complex mapping, and GNoME materials discovery.

See the benchmarks
05 · Humanity

Post-training, safety & alignment

SFT, DPO, and GRPO reinforcement learning. Analyzing alignment failures, Sleeper Agents, sycophancy, labor impacts, and macroeconomic capex trends.

Analyze the impact
06 · AGI & ASI

AGI timelines & new scaling laws

We map every prediction from Hassabis, Amodei, Legg, and Altman. Inspect Chinchilla scaling limits and the pivot to test-time compute scaling.

Read the timelines
07 · Play

AI in games & virtual worlds

David Silver's reinforcement learning milestones from AlphaGo to AlphaStar. Explore Genie 3 real-world engines and Factorio long-horizon agents.

Step into the simulation
08 · Next Decade

Horizon tech & physical robotics

Quantum computing advantage roadmaps (Willow, Starling), neuromorphic chips, DNA storage, and humanoid robotics (Optimus Gen 3, Figure 02).

See the roadmap
09 · The Grid

Infrastructure & nuclear deals

Powering intelligence at gigawatt scale: Meta's 5 GW Hyperion, xAI Memphis, and nuclear PPAs (Three Mile Island, Kairos SMR, Amazon Talen).

Inspect the grid
10 · Geopolitics

The Chip Wars & trade controls

Taiwan's TSMC chokepoint, ASML's High-NA EUV lithography monopoly, U.S. export controls on advanced compute, and China's domestic Ascend ecosystem.

Explore the supply chain
11 · The Weights

Open weights vs closed APIs

The commercial and philosophical weights war. Meta's Llama 4 Scout, Mistral Small 4, Gemma 4, and the regulatory impacts of California's legislation.

Compare the weights
12 · The Data Wall

Data depletion & synthetic futures

Internet data exhaustion timelines. How test-time compute, RLAIF self-play, and sandbox simulations bypass the physical data wall.

Study the new scaling laws
Appendix · Daily Practice

Daily AI Fluency: learn to practice

Six practical habits to build AI literacy: prompting, agent execution, daily utilization, and keeping up with practitioners over pundits.

Build your AI practice
What the builders are saying
We are past the event horizon; the takeoff has started. Humanity is close to building digital superintelligence.
Sam Altman · OpenAI · May 2026
We are near the end of the exponential — a few years away from a country of geniuses in a data center.
Dario Amodei · Anthropic · Feb 2026
We are standing in the foothills of the singularity. Society has only a few years left to prepare.
Demis Hassabis · Google DeepMind · May 2026
Digital superintelligence is coming. xAI is positioning Colossus 2 and Grok 5 to achieve true AGI.
Elon Musk · xAI · 2026
LLMs will not lead to AGI. Objective-driven world models are the true path forward.
Yann LeCun · AMI Labs · Feb 2026
We are not here to win a price war. We are here to push the boundary of intelligence, raising $7B to build open weights.
Liang Wenfeng · DeepSeek · June 2026