Another year of tracking emerging technology comes to a close. Looking back at 2025, certain themes dominated: AI maturation, energy transition tensions, biotech advances, and the ongoing search for the next big thing. Here’s my assessment of what mattered.

The AI Story

2025 was the year AI stopped being future technology and became present reality. Not in the “AGI is here” sense the breathless headlines suggested, but in the “every enterprise software now has AI features” sense.

Model capabilities plateaued (sort of). After dramatic improvement in 2023-2024, the capability curve flattened. GPT-5 and equivalent models were better than their predecessors, but incrementally so. The age of “wait six months for dramatically better models” seems to be ending.

Enterprise deployment accelerated. The more important story than model improvements was deployment. Companies that had been experimenting moved to production. AI-assisted customer service, document processing, code generation, and internal knowledge management went from pilots to operations.

The evaluation problem became acute. As deployment scaled, so did the challenge of measuring whether AI systems were actually working. Organizations struggled to assess accuracy, catch regressions, and understand failure modes.

Regulation started biting. The EU AI Act’s requirements began affecting real decisions. Organizations had to classify their AI systems, document their approaches, and in some cases modify or abandon projects.

Agent hype met reality. The “autonomous agent” vision remained largely unfulfilled. What worked were narrow, supervised agents handling specific tasks. The fully autonomous AI coworker remained science fiction.

Energy and Climate

The energy transition remained contested but continued:

Nuclear resurgence. The tech sector’s embrace of nuclear power was the surprise of the year. Data center electricity demand created customers willing to pay premium prices for reliable, zero-carbon power.

Grid strain became visible. Electricity demand growth from AI, EVs, and electrification hit grid capacity limits in some regions. The infrastructure challenge became harder to ignore.

Climate tech investment matured. Less speculative capital, more project finance. The sector looked more like traditional energy and less like venture-backed tech.

Green hydrogen advanced slowly. Demonstration projects progressed but commercial-scale green hydrogen remained elusive. The economics remained challenging without policy support.

Biotech and Health

Biotech AI integration deepened. The combination of AI and biotech moved from buzzword to standard practice. Drug discovery pipelines increasingly incorporated AI tools for molecule design and target identification.

GLP-1 drugs dominated. The obesity drug market exploded, creating questions about long-term health system impacts and manufacturing capacity.

Gene therapy expanded. More gene therapies reached market, though pricing and access challenges persisted.

Synthetic biology commercialized. Products made through engineered biology moved from novelty to meaningful market presence, particularly in food ingredients and specialty materials.

Computing and Infrastructure

NVIDIA’s dominance continued. Despite much discussion of competition, NVIDIA remained the default choice for AI compute. Alternatives made progress but didn’t fundamentally change the market.

Edge AI deployment grew. More AI moved from cloud to edge - into factories, vehicles, and devices. The infrastructure for distributed AI intelligence expanded.

Quantum remained pre-commercial. Progress continued on error correction and qubit counts, but practical quantum advantage for business problems remained future-tense.

Space infrastructure matured. The commercial space ecosystem became more clearly valuable for terrestrial applications - communications, Earth observation, positioning.

The Things That Didn’t Happen

What I expected that didn’t materialize:

Consumer AI hardware. Various AI-focused consumer devices launched but none found mass market traction. The smartphone remained the AI interface.

Web3 revival. Specific applications (stablecoins, tokenization) progressed, but the broader Web3 ecosystem didn’t return to 2021-era enthusiasm.

Autonomous vehicle breakthrough. Robotaxis expanded service areas incrementally but the promised transportation revolution remained incremental.

Major AI regulation backlash. Some predicted that AI regulation would kill innovation. It didn’t. Companies adapted, and the pace of deployment continued.

What It Means for 2026

Looking ahead, several dynamics seem likely:

AI deployment deepens but problems emerge. As more AI systems run in production, more things will go wrong. Expect at least one major public AI failure that accelerates governance discussions.

The energy-AI nexus tightens. Data center electricity demand will increasingly drive energy policy and investment. The industries will become more intertwined.

Biotech-AI convergence accelerates. The most interesting innovation in both fields may happen at their intersection.

Climate tech faces political headwinds in some regions. Policy support isn’t guaranteed. Technologies that work economically without subsidy will be more robust.

Commoditization pressures intensify. AI capabilities that are differentiated today will be table stakes tomorrow. The competition shifts to who can deploy effectively.

Personal Takeaways

What I learned this year:

Deployment is harder than development. The technology often exists before organizations can effectively use it. The bottleneck is increasingly organizational, not technical.

Patience remains a competitive advantage. In a world of hype cycles, those who can take a long view have advantages.

Integration matters more than novelty. The most valuable innovations aren’t necessarily the most technically impressive - they’re the ones that fit into existing workflows and systems.

Humility is appropriate. I got things wrong this year. Everyone tracking emerging technology does. The goal isn’t perfect prediction - it’s useful orientation that can be updated as evidence accumulates.

2025 was another year of technology progress that felt both faster and slower than expected - faster in specific areas, slower in the grand narrative. That pattern will likely continue.

See you in 2026.