Making predictions is risky. You’re wrong in public, and people remember. But I find the exercise useful for crystallizing thinking, so here goes.

Five specific predictions for emerging technology in 2026, along with my reasoning and what to watch.

Prediction 1: AI Agent Deployment Hits Real Scale, But Not How You Expect

The prediction: 2026 will be the year AI agents move from pilots to production at enterprise scale. But the successful deployments won’t be autonomous agents doing open-ended work - they’ll be tightly constrained agents handling specific, well-defined workflows with robust human oversight.

The reasoning: The technology for capable AI agents exists. What’s lagged is the organizational infrastructure to deploy them safely. In 2026, we’ll see that infrastructure mature: better monitoring, better guardrails, better integration with existing systems, clearer regulatory guidance.

The companies that succeed won’t be the ones promising “autonomous AI workers.” They’ll be the ones offering “AI that handles 80% of your routine work while humans handle exceptions.”

What to watch: Enterprise deployments with published metrics on throughput and error rates. If we see companies sharing specific numbers about agent performance in production, the real scaling has begun.

Confidence level: High. The pieces are in place; it’s a matter of maturation.

Prediction 2: The First Major AI-Related Enterprise Disaster

The prediction: A major company will suffer a significant public failure - data breach, financial loss, regulatory penalty, or reputational crisis - directly attributable to AI deployment. This will accelerate AI governance discussions and create market opportunity for AI risk management.

The reasoning: AI is being deployed faster than risk management practices are developing. Somewhere, someone is deploying AI without adequate testing, monitoring, or human oversight. As deployment scales, so does the probability of something going badly wrong.

We’ve had small incidents. We haven’t had a headline-grabbing disaster yet. Statistics suggest we will.

What to watch: Early warning signs in sectors with aggressive AI deployment and high stakes: financial services, healthcare, customer-facing applications.

Confidence level: Medium-high. The disaster is likely; timing is uncertain.

Prediction 3: Quantum Computing Hype Returns

The prediction: After a quiet period, quantum computing will experience another hype cycle in 2026, driven by specific technical milestones around error correction and early demonstrations of quantum advantage for practical problems.

The reasoning: The serious players (IBM, Google, IonQ, Quantinuum) have roadmaps that converge on meaningful milestones in the 2025-2027 timeframe. If any of them hit their targets - even partially - it’ll reignite excitement.

The hype won’t be justified by near-term commercial applications (still years away), but the market doesn’t always wait for commercial reality.

What to watch: Papers demonstrating quantum error correction below threshold. Any demonstration of practical quantum advantage (not just benchmark problems).

Confidence level: Medium. Contingent on technical milestones that may or may not be achieved.

Prediction 4: Climate Tech’s Second Wave

The prediction: Climate tech investing will surge again, but with different character than the 2021 wave. Less consumer-facing, more industrial. Less SPAC-driven speculation, more patient capital backing hard tech.

The reasoning: The policy environment (IRA, EU Green Deal equivalents) provides durable tailwinds. The technologies that were speculative in 2021 (green hydrogen, long-duration storage, sustainable aviation fuel) have advanced enough for larger-scale deployment.

Meanwhile, the first wave’s excesses have cleared - failed SPACs, down rounds, bankruptcies. What’s left is more serious.

What to watch: Institutional investor allocations to climate tech. Project finance deals for climate tech manufacturing facilities.

Confidence level: Medium-high. The fundamental drivers are strong.

Prediction 5: Enterprise AI Becomes Boring (In a Good Way)

The prediction: By end of 2026, basic AI capabilities (document processing, search enhancement, workflow automation) will be unremarkable features of enterprise software, not selling points. AI will stop being a differentiator and become expected table stakes.

The reasoning: Every enterprise software vendor is adding AI capabilities. Most will ship something adequate in 2026. When everyone has “AI-powered” features, AI ceases to be remarkable.

This is healthy commoditization. The market will shift from “do you have AI?” to “how well does your AI work for my specific needs?”

What to watch: Enterprise software marketing that emphasizes something other than AI. When vendors stop leading with “AI-powered,” the commoditization is complete.

Confidence level: High. Already underway; question is timeline.

How to Use Predictions

Predictions aren’t gospel. They’re frameworks for thinking about the future and identifying what to watch.

If I’m wrong about these, the interesting question is why. What did I miss? What assumptions failed?

If I’m right, the next question is what to do about it. How does an AI agent deployment wave affect your strategy? How does enterprise AI commoditization change your competitive position? What does a climate tech resurgence mean for your portfolio?

Predictions have value not as certainties but as structured thinking about uncertainty. Use them to identify the signals that would confirm or disconfirm specific futures, then watch for those signals.

I’ll revisit these in twelve months and see how we did. Public accountability, as always.