Technology Timeline

    From Electricity to Artificial Intelligence

    This timeline traces the complete arc of intelligent machines — from Michael Faraday's electric motor in 1821 to the AI agent economy of 2026. Every major milestone that led to the current AI era is represented, spanning electricity, computing hardware, software infrastructure, robotics, and artificial intelligence.

    Each dot represents a milestone. Dot size reflects historical significance. Colours indicate category: AI & Machine Learning, Robotics, Chips & Hardware, Energy & Physics, Software & Internet, and Commercial Milestones.

    Hover over any dot to see the event description. Brighter dots with a pointer cursor link to full encyclopedia entries — click to read the complete analysis. On mobile, scroll the vertical list to explore the full history.

    The timeline begins in 1821 with Faraday's electric motor — the foundational enabling technology without which none of the subsequent computing, semiconductor, or AI revolutions would have been possible. Every modern data centre still runs on Faraday's principle. Reading left to right reveals the technological lineage: electricity unlocks computation, computation unlocks software, software unlocks intelligence. Click any linked event to read the full encyclopedia entry on that milestone and trace the chain of dependencies forward and backward through the Encyclopedia.

    1990s
    2000s
    2010s
    2011D-Wave Annealer Ships
    D-Wave Systems delivers its first commercial quantum annealing system to Lockheed Martin, marking the first commercial sale of a quantum computing system despite ongoing debate about whether the device demonstrates genuine quantum speedup.
    2012AlexNet Deep Learning
    AlexNet wins the ImageNet competition by a massive margin using deep convolutional neural networks, launching the deep learning revolution.
    2013Boston Dynamics Atlas
    Boston Dynamics unveils Atlas, a humanoid robot capable of dynamic balance and complex physical tasks.
    2014Amazon Warehouse Bots
    Amazon deploys Kiva Systems robots across its fulfilment centres, transforming warehouse automation at scale.
    2014DeepMind Acquired
    Google acquires DeepMind for $500 million, signalling that AI research is a strategic priority for the world's largest technology companies.
    2015OpenAI Founded
    Elon Musk, Sam Altman and others found OpenAI as a nonprofit AI safety research organisation — it later becomes the most consequential AI company in history.
    2016AlphaGo v World Champ
    DeepMind's AlphaGo defeats world Go champion Lee Sedol — a game considered far beyond computers — using deep reinforcement learning.
    2017Google Transformer
    Google researchers publish Attention Is All You Need, introducing the transformer architecture that underpins every major AI system since.
    2018GPT-1 Released
    OpenAI releases GPT-1, the first large language model based on the transformer architecture, demonstrating text generation at scale.
    2019Sycamore Supremacy
    Google announces its 53-qubit Sycamore processor completed a specific computation in 200 seconds that would take classical supercomputers 10,000 years, claiming quantum supremacy. IBM contests the claim, arguing classical optimisation could narrow the gap.
    2020s
    2020GPT-3 Released
    OpenAI releases GPT-3 with 175 billion parameters — an order of magnitude larger than any previous model — demonstrating emergent capabilities.
    2020Waymo Robotaxis
    Waymo launches the first fully driverless commercial robotaxi service in Phoenix, Arizona — no safety driver.
    2021DALL-E Image Gen
    OpenAI's DALL-E demonstrates AI image generation from text prompts, opening generative AI to visual media.
    2022ChatGPT Launch
    OpenAI launches ChatGPT — it reaches 100 million users in 60 days, the fastest product adoption in history, triggering the mainstream AI era.
    2023Nvidia $1 Trillion
    Nvidia crosses $1 trillion market capitalisation as AI chip demand drives one of the fastest wealth creation events in stock market history.
    2023GPT-4 Multimodal
    OpenAI releases GPT-4 with multimodal capabilities, passing professional exams and approaching human performance on complex reasoning tasks.
    2023Qubotics Airport Bots
    Qubotics applies Quadratic Unconstrained Binary Optimisation to automate ground baggage handling at airports, representing one of the first commercial deployments of quantum-inspired optimisation in physical robotics systems. The approach uses QUBO mathematical frameworks compatible with both quantum annealers and classical hardware.
    2023Figure Humanoid Robot
    Figure AI demonstrates Figure 01 performing autonomous tasks in a warehouse, backed by BMW and Microsoft.
    2024Google Willow Chip
    Google's 105-qubit Willow processor achieves below-threshold quantum error correction, where adding more qubits reduces rather than increases overall error rates. The chip solves a benchmark in 5 minutes that would take classical computers 10 septillion years.
    2024NIST Post-Quantum Std
    The National Institute of Standards and Technology publishes the first post-quantum cryptographic standards, selecting algorithms resistant to both classical and quantum attacks. Enterprises begin mandatory migration away from RSA and ECC encryption.
    2024Tesla Optimus
    Tesla deploys Optimus humanoid robots in its own factories, targeting production of millions of units per year.
    2024AI Agents Emerge
    AI agents — systems that autonomously plan and execute multi-step tasks — move from research to commercial deployment across enterprise software.
    2024AI Coding Mainstream
    AI coding assistants including Cursor and GitHub Copilot are adopted by the majority of professional developers, beginning the transformation of software engineering.
    2024Nvidia AI Dominance
    Nvidia briefly becomes the most valuable company in the world as AI chip demand drives a historic stock rally.
    2024AI Investment Surge
    Global AI investment exceeds $100 billion in a single year as hyperscalers commit to building AI infrastructure at unprecedented scale.
    2025EU AI Act Enforcement
    The EU AI Act's obligations for general-purpose AI models begin applying from August 2025 — the world's first comprehensive AI regulation enters enforcement, reshaping how AI companies operate globally.
    2025DeepSeek V3.2
    DeepSeek V3.2 achieves gold medal at IMO 2025 and is released free under Apache 2.0, proving Chinese labs can match US frontier AI despite chip export restrictions — at 10x lower cost.
    2025Claude 4 Released
    Anthropic releases Claude 4 with Opus 4 becoming the world-leading coding model, achieving 72.5% on SWE-bench and demonstrating 30+ hour autonomous operation capability for complex agentic tasks.
    2025AI Model Race Nov 25
    November 2025 becomes the most competitive month in AI history — GPT-5.1, Gemini 3 Pro, Grok 4.1, and Claude Opus 4.5 all release within six days, each briefly claiming the top benchmark position.
    2025AI Outperforms Doctors
    Microsoft's AI Diagnostic Orchestrator solves complex medical cases with 85.5% accuracy, compared to 20% for experienced physicians — marking the first clear domain where AI clinical performance exceeds human specialists.
    2025Meta Llama Multimodal
    Meta releases Llama 4 — the first natively multimodal open-source model family, understanding text, images, and video simultaneously, making frontier multimodal AI freely available to all.
    2025GPT-5 Released
    OpenAI releases GPT-5 in August 2025, closing the gap with human expert performance on professional benchmarks and introducing adaptive reasoning that adjusts compute based on task complexity.
    2025Humanoids at Mercedes
    Apptronik's Apollo humanoid robot deploys in Mercedes-Benz plants across Europe for intralogistics, alongside UBTECH Walker S2 beginning mass production with orders exceeding $112 million.
    2025MCP — The USB-C for AI
    Anthropic launches the Model Context Protocol (MCP), a standard interface letting any AI model connect to external tools, databases, and APIs. Every major AI provider adopts it within months.
    2025MS Topological Qubits
    Microsoft announces its topological qubit architecture achieves the error rates required for fault-tolerant quantum computation, representing a potential alternative path to large-scale quantum computing using Majorana fermions rather than superconducting qubits.
    2025Digit Deployment
    Agility Robotics' Digit becomes the first humanoid robot in commercial warehouse deployment, with Amazon actively using it in fulfilment centres and Digit passing 100,000 totes moved at a GXO facility.
    2025AI Agent Protocols
    Google releases Agent-to-Agent (A2A) protocol enabling AI agents to communicate directly with each other, alongside IBM's ACP — establishing the infrastructure standards for multi-agent AI systems.
    2025Stanford AI Index
    Stanford HAI's AI Index reveals AI capability growth has nearly doubled in pace — coding benchmark scores jumped from 60% to near 100% in a single year, and frontier models now exceed 50% on Humanity's Last Exam.
    2025DeepSeek R1
    Chinese lab DeepSeek releases R1 — an open-source reasoning model matching OpenAI o1, trained for just $5.6M. The release triggers a global debate about AI compute requirements and sends Nvidia stock down 17% in a single day.
    2025AI Coding Adoption
    GitHub reports developers merged 43 million pull requests monthly in 2025 — a 23% increase — as AI coding tools reach majority adoption among professional developers globally.
    2025Unitree R1 at $5900
    Chinese robotics company Unitree releases the R1 humanoid robot at $5,900 — the first sub-$10,000 capable humanoid, deploying in BYD and Geely EV factories and signalling that humanoid robots are approaching consumer price points.
    2025Figure 03 at BMW
    Figure AI releases Figure 03 in October 2025 — in an 11-month pilot at BMW's Spartanburg plant, Figure robots contribute to production of over 30,000 vehicles, the most significant humanoid-automotive integration to date.
    2026MCP 97M Installs
    Anthropic's Model Context Protocol crosses 97 million installs and is contributed to the Linux Foundation under open governance, becoming the universal standard for AI agent tool integration.
    2026NASA Mars Navigation
    NASA's Perseverance rover completes the first Mars drives ever planned by artificial intelligence using Claude vision-language models, autonomously navigating 456 meters — replacing 28 years of manual human planning.
    2026Agentic AI Enterprise
    Multi-agent AI systems move from lab to production across enterprise software — AI agents that autonomously plan, execute, and verify multi-step workflows become standard in finance, legal, and software development.
    2026IBM Quantum Advantage
    IBM demonstrates the first quantum computer outperforming all classical methods on a real-world problem — the long-anticipated "quantum advantage" milestone arrives, with implications for drug discovery, materials science, and cryptography.
    2027First Humanoid RobotProjected
    Figure, 1X, and Agility Robotics reach combined deployments of 10,000+ humanoid units in warehouse and manufacturing environments — the threshold at which unit economics begin to compete with human labour for structured physical tasks.
    2027IBM 10k Qubits TargetProjected
    IBM roadmap milestone targeting 10,000 physical qubits, representing an order-of-magnitude increase from the 1,121-qubit Condor system. At projected error correction ratios, this would yield approximately 10 logical qubits — still far short of cryptographically relevant computation but a significant engineering milestone.
    2027TSMC 2nm ProductionProjected
    TSMC begins volume production on its N2 node, delivering approximately 15% performance improvement and 25% power reduction versus N3. Apple and NVIDIA are expected as launch customers, setting the baseline for next-generation AI accelerators.
    2028ITER First PlasmaProjected
    ITER achieves first plasma in its Cadarache tokamak, validating the engineering design for a net-energy-producing fusion reactor. First plasma is not energy breakeven — that milestone is targeted for the early 2030s — but confirms the construction programme.
    2028AI Agents in WorkflowProjected
    Enterprise AI agents begin managing multi-step workflows — procurement, scheduling, customer support escalations — without human approval at each step. Marks the transition from AI as a tool to AI as a participant in business operations.
    2029US Fed PQC DeadlineProjected
    US Office of Management and Budget deadline requiring all federal agencies to complete migration to NIST-standardised post-quantum cryptographic algorithms. Mandated following NIST finalising PQC standards in 2024, this represents the largest coordinated cryptographic infrastructure upgrade in US government history.
    2030s
    2030Robotaxis MainstreamProjected
    Robotaxi services operate commercially in 20+ major cities globally without safety drivers. Waymo, Baidu Apollo, and Zoox reach this threshold via regulatory approvals in US, China, and EU markets.
    2030Fault-Tolerant QuantumProjected
    Industry and academic consensus projects that leading quantum computing systems will approach the fault-tolerant threshold around 2030, where logical qubit counts become sufficient for narrow but genuine commercial applications in molecular simulation and financial optimisation. Timeline carries significant uncertainty in both directions.
    2030AGI CapabilityProjected
    AI systems begin passing comprehensive general competency evaluations across domains — science, mathematics, coding, planning, and social reasoning — at human expert level simultaneously. Whether this constitutes AGI remains contested, but the benchmark threshold is crossed.
    2033Fault-Tolerant Quantum Computing ThresholdProjected
    IBM or Google achieves a fault-tolerant logical qubit array sufficient to run Shor's algorithm against small RSA keys — the milestone that begins the post-quantum cryptography transition urgency for financial systems.
    2033IBM 100k Qubits TargetProjected
    Published IBM quantum roadmap targets 100,000 physical qubits by 2033 through modular quantum processor architecture. After error correction overhead, this represents approximately 100 logical qubits — a significant milestone but still roughly 40x short of the logical qubit count required to threaten RSA-2048 encryption via Shor's algorithm.
    2035Human-Level RoboticsProjected
    Robotic systems achieve human-level dexterity and adaptability in unstructured physical environments — kitchens, construction sites, elder care settings — validated by standardised manipulation benchmarks. This milestone marks the point at which robotics can address the full scope of physical labour.
    2035Quantum Drug DiscoveryProjected
    Industry analyst projections place the first commercially validated quantum advantage in pharmaceutical molecular simulation around 2035, when fault-tolerant quantum systems may be capable of simulating medium-complexity protein interactions with accuracy beyond classical approximation methods. Multiple major pharmaceutical companies are positioning quantum pipelines in anticipation.
    2035Bitcoin Q-Day WindowProjected
    Optimistic end of the 15-25 year estimated range for cryptographically relevant quantum computers capable of running Shor's algorithm against Bitcoin's secp256k1 elliptic curve cryptography. Requires millions of physical qubits yielding approximately 4,000 logical qubits — roughly 40x beyond IBM's 2033 roadmap target. Most analysts place Q-Day between 2040-2050 under current trajectories.

    Key Takeaways

    • ▸The arc of intelligent machines starts with Faraday's 1821 electric motor — every modern data centre still runs on that principle.
    • ▸Electricity unlocks computation, computation unlocks software, software unlocks intelligence — each layer makes the next economically viable.
    • ▸Six tracks are colour-coded: AI & Machine Learning, Robotics, Chips & Hardware, Energy & Physics, Software & Internet, and Commercial Milestones.
    • ▸Dot size reflects historical significance; linked dots open the full Encyclopedia entry for that milestone.
    • ▸The 2010s ignition (deep learning, GPUs, cloud, mobile data) compounds into the 2020s agent economy currently tracked on the Indicators page.

    Frequently Asked Questions

    Why does the timeline start in 1821?

    1821 is the year Michael Faraday demonstrated the first electric motor — the foundational enabling technology without which none of the subsequent computing, semiconductor or AI revolutions would have been possible. Every modern data centre, GPU and robot still runs on Faraday's principle, so the chain of dependencies for intelligent machines begins there.

    How are milestones selected and ranked?

    Each milestone is included because it materially changed what came next — a new enabling capability, a step-change in cost or scale, or a commercial deployment that pulled the rest of the industry forward. Significance, encoded as dot size, reflects downstream impact: how many later milestones depended on this one to exist.

    What do the colours mean?

    Colour encodes category so you can trace lineages at a glance: AI & Machine Learning, Robotics, Chips & Hardware, Energy & Physics, Software & Internet, and Commercial Milestones. Reading left to right shows how breakthroughs in one track unlock progress in the next.

    How do I read the linked events?

    Brighter dots with a pointer cursor link to a full Encyclopedia entry on that milestone. Click through to read the long-form analysis, the chain of dependencies forward and backward, and the connections to current Indicators and Analysis briefings.

    Where does the timeline end?

    The visible end of the timeline is the present day — currently the AI agent economy of 2026. The Indicators page tracks the live readings that extend the timeline forward in real time, and the Analysis section interprets what new milestones mean for the broader investment thesis.

    Related analysis