Moore’s Law is the observation that computing performance tends to double every two years. It held true for decades, yet in 2023 was shattered with hardware and software performance achieving 10x gains in just six months. In 1956, one megabyte of data storage cost $9,000 (equivalent to $87,000 in 2025 adjusted for inflation), but had dropped to just $0.00001 per megabyte in 2025.

NVIDIA is the leader is AI ‘chips’ (GPUs and IPUs). In 2025 their single integrated ‘chip,’ selling for US$250,000 and with 35,000 components, could achieve the performance of a previous generation data center. The performance gains are staggering, and the cost of computation in 2025 was roughly one hundred-millionth of what it was in the 1970s. In 1984, the cost of performing 1 gigaflop (one billion floating-point operations per second) was $18,700,000, equivalent to $47,00,000 million in 2025 when it had dropped to just $0.01 per gigaflop—a reduction in computational cost of more than 99.99%.

Beyond the price to capability equation, raw intelligence has been exponentially accelerating. In 2016, AlphaGO from DeepMind (now part of Google) won the most complex game in the world, GO! It easily beat Lee Sedol, the eighteen-times human world champion. In doing so, it shocked the world with self-learned creativity and new strategies never seen previously. Move 37 in game two of the five game contest is AI legend and demonstrated the power of large language model neural networks… and self-generated creativity.

AI transformer technology was first publicly released by Google in 2017. Generative Pretrained Transformers (GPTs) utilize Large Language Models (LLMs) that can be prompted to generate unique content. The best-known leader in this field was OpenAI with ChatGPT. These systems were originally trained to create written content before advancing to also generate voice, images, art, videos, multi-character podcasts, websites, digital [human] twins, software coding and programs, and more. Symbolic classical Artificial Intelligence (AI) evolved from binary logic and algorithms, to also mimic neural networks found in nature, specifically the human brain. AI then further evolved from generative to agentic, creating ‘digital labor’ in the form of ‘virtual agents’ to execute tasks previously done by humans, ultimately with human-like conversational interfaces.

In 2023, Google’s DeepMind, which had previously dominated the games of Go and chess, was tasked with structurally replicating all the known proteins in nature. This task would have taken human PhD postgraduate students 1,000,000,000 hours to complete. Instead, the Alpha AI was tasked with understanding and executing all 200,000,000 known protein sequences in nature—the amino acids which form the building blocks in molecular biology. AlphaFold achieved the task of accurately ‘folding’ (predicting and generating 3D representations) of every known protein in nature in just one year, and without any human assistance. The research was later gifted to the scientific community for the purpose of creating synthetic vaccines and myriad other applications. In 2025, DeepMind launched AlphaEvolve for evolutionary algorithms to autonomously self-improve software models and code at unprecedented speed and scale.

Quantum computing represents the next revolutionary breakthrough in science and technology. Despite its potential to operate 1,000,000 times faster than a supercomputer, it is hindered by qubit instability. But even with the instability problem, in 2024 the world's most powerful supercomputer was estimated to require 45 years to process calculations that an advanced quantum computer can achieve in mere seconds. Stable quantum computing has the potential to overcome all current encryption security measures while heralding a new era of machine super-intelligence. Experts agree that encryption security protections of IT systems will eventually be dismantled with the advances of quantum computing.

In 2025, Grok from xAI surpassed human-level intelligence in physics, mathematics and language. Other frontier labs (ChatGPT, Gemini, DeepMind, Claude, and others) also set astounding intelligence records. AI now passes PhD-level exams with 100% scores. Humanity’s Last Exam is the most difficult intelligence test devised by humans and is the benchmark test for AI labs that are relentlessly improving.

The AI race is being driven by commercial, military, political, and scientific objectives… but with politicians and policy hopelessly lagging behind. Expert consensus is that AGI (Artificial General Intelligence) is immanent, and that super-intelligence will soon follow.  Some claim it has already happened but with humanity being unaware. Conversational and visual AI is becoming hyper-realistic, with people regularly fooled in their interactions and by fake content. Conversation AI is becoming embraced as a philosophical companion that shapes opinions and beliefs while addicting the human user. Embodied AI is achieving breakthrough advances with humanoid robots and myriad specialist machines. Yet all of this advancement occurs while safety takes a back seat in the AI race by governments and corporations.