From Agentic AI to Fusion Energy: 4 Forces Reshaping 2026-2028
Methodology:
Three years ago, I published "From Space Travel to Metaverse, 4 Trends that will shape 2022" using a straightforward approach: evaluate the foundational shifts materializing in real time, then apply a forward-looking lens to see where they lead. I'm using the same methodology here, but with one critical change based on watching how my 2022 predictions actually played out.
I'm now much more focused on deployment timelines, regulatory clearance, and actual customer commitments rather than technical capability alone, because I learned the hard way that technology working in a lab is radically different from technology shipping at scale.
From evaluating the current landscape, I've distilled the list down to Top-4, listed below in ascending order of impact.
Scorecard: What I Got Right and Wrong in 2022
Before looking forward, let's be honest about what actually happened with my 2022 calls, because understanding the gap between prediction and reality is how you get better at this.
Space Travel for the masses:
I wrote that private space flights would accelerate in 2022, with SpaceX's Starship leading the charge toward something resembling mass adoption. I was directionally right about the trend but badly wrong about the speed and timeline, which is probably the most humbling kind of wrong because the direction matters but the timing matters more.
Virgin Galactic flew its last commercial flight in June 2024 and suspended operations for two years while building Delta-class vehicles, which means the company now targets resuming customer flights in fall 2026, essentially putting us back where we would have been in 2022 if timelines had held. Blue Origin completed ten suborbital missions through 2025, but the scale remains tiny compared to what "mass adoption" would require, and the volume advantage that matters for cost reduction simply hasn't materialized yet. The space tourism market is projected to grow from $1.1 billion in 2025 to $18.4 billion by 2033, which represents strong growth at 38.6% CAGR, but that trajectory confirms this is fundamentally a decade-long infrastructure build, not the 2-3 year acceleration I was predicting.
What I underestimated was the compounding complexity of manufacturing capacity constraints, the regulatory and safety verification requirements around human-rating systems, and how slowly technical maturity translates into operational deployment when human safety is the primary concern. Hardware timelines are brutal in ways that software timelines aren't, and space tourism proved no exception to that rule.
Crypto, Decentralized Finance (DeFi) and Decentralized Autonomous Organizations (DAO):
I predicted crypto and DeFi would go mainstream in 2022, which turned out to be right in direction but wrong on speed in much the same way the space tourism prediction was wrong on speed. By 2025, the pieces finally fell into place after years of regulatory back-and-forth that nobody predicted would take quite so long. Stablecoin supply hit record levels, with institutional issuers like BlackRock's USDTB and PayPal's PYUSD attracting billions in capital flows, and Stripe integrated stablecoin settlement directly into its payment infrastructure, which was the kind of mainstream adoption signal that actually matters because it means actual companies are building real products on top of it.
The regulatory environment shifted from outright hostility to cautious acceptance, creating the clarity institutions needed to participate without risking regulatory blowback, and that shift mattered more than the technology itself because institutions don't move until they have clarity.
According to Grayscale's 2026 Digital Asset Outlook, we're entering the "dawn of the institutional era" for crypto, with DeFi lending protocols like Aave and Morpho achieving meaningful growth and stablecoins becoming core infrastructure for cross-border payments in ways that actually produce ROI.
DeFi moved from speculative technology to functional infrastructure, but it took three full years instead of one, which taught me an important lesson about the gap between what I predicted and what actually happened: the technology can be ready long before the regulatory and institutional framework catches up, and that lag time matters because it determines whether adoption is accelerating or just delayed.
Metaverse & NFT (non-fungible Tokens):
I was flat wrong about this one, or at least wrong about the timeline and the path adoption would actually take, which is a more useful kind of wrong because it forces you to examine the assumptions behind the prediction rather than just dismissing it as a bad call.
In 2022, I cited Wall Street projections claiming the metaverse would become a trillion-dollar market and quoted Morgan Stanley's forecast that NFTs and social gaming could reach $50 billion by 2030 for luxury brands alone, which felt credible at the time because the hype was everywhere and major institutions were making those calls. What actually happened was a textbook speculative bubble followed by a painful correction that separated the narratives from the actual business models that could sustain themselves.
NFT market revenue peaked at $1.58 billion in 2022, then crashed and stabilized at a range of $600-700 million by 2024-2025, which tells you something important about market psychology: the number of NFT sales increased nearly 80% in the first half of 2025, but total dollar volume declined 4.6%, which is the signature of speculation washing out and being replaced by actual utility-based demand. More activity at lower valuations means the market is hunting for real use cases rather than momentum trades.
The consumer metaverse hype collapsed almost entirely, though what survived is a smaller, more practical market focused on enterprise use cases and specific gaming applications where digital ownership creates tangible value that justifies the cost rather than just narrative appeal. The metaverse NFT market is still projected to grow from $12.5 billion in 2025 toward higher numbers by 2033, but we're nowhere close to the explosive trajectory that was predicted in 2021-2022 when everything felt like it was accelerating exponentially.
What I got wrong was allowing narratives to influence my thinking more than fundamentals should have, because prices peaked before infrastructure was ready and the market ran ahead of actual utility. When reality caught up, valuations had to reset hard, which is painful but necessary, and now I look for paying customers and measurable use cases rather than projections and pitch decks that sound good until they hit reality.
Privacy, Transparency, Governance and Accountability:
This prediction was more observation than forecast, and it's aged well precisely because it was structural rather than hype-driven, which means it kept grinding forward at a steady pace regardless of market sentiment or investor enthusiasm. The EU AI Act's compliance timeline begins in August 2026, regulatory frameworks continue maturing across jurisdictions at the pace these things naturally mature, and organizations now treat governance as part of the product rather than an afterthought. This trend didn't accelerate or crash like the others because it was driven by regulation and institutional necessity rather than speculation.
What Continues Beyond 2025
Two of my 2022 predictions have real staying power moving into 2026-2028, though they're arriving on different timelines and with important modifications to how I was thinking about them.
DeFi has transitioned from bubble to infrastructure. Stablecoins now function as the monetary base for digital payments and remittances in ways that actually work reliably at scale, and the technology has proven itself after years of stress testing and real-world operation. Adoption will accelerate as regulatory frameworks solidify and more institutions integrate crypto rails into treasury operations and cross-border settlement, which is starting to happen now. By 2028, I expect DeFi to be standard infrastructure for mid-market companies, not just crypto-native firms, because the operational advantages over traditional banking will become obvious to CFOs running those companies.
Space tourism will scale slowly but steadily through the late 2020s. Virgin Galactic targeting commercial operations in fall 2026 is concrete evidence that the industry is moving forward, though it's clear this is a long-cycle capital-intensive build that requires new manufacturing capacity, regulatory maturity, and customer education at every step. By 2028, space tourism will still be a luxury experience for high-net-worth individuals rather than mass-market travel, and that's fine because that's where the revenue actually is right now, so this is a case where the slower timeline makes the business model more sustainable rather than less.
Four New Forces Shaping 2026-2028; The Next 3 Year Predictions & Why They Matter?
Agentic AI: When Pilots Hit Production Reality
Most of the coverage around agentic AI focuses on capabilities and demos, which is interesting but misses the actual story of whether this technology is shipping into real businesses at scale. Here's the statistic that actually matters and should change how you think about the space: 38% of organizations are piloting agentic AI, but only 11% have deployed it in production, which is a gap so massive that it's not shrinking as fast as anyone expected and will only narrow through a painful sorting process.
I've watched this pattern repeat across industries in painful detail. Companies bolt agentic AI onto existing broken processes instead of redesigning workflows end-to-end, treating it as a cost-cutting tool rather than a strategic capability that changes how work fundamentally gets done. The result is predictable: systems that look great in controlled demos but fail catastrophically at scale when they hit legacy infrastructure, inconsistent APIs, and missing governance frameworks that nobody built because they assumed the AI would just work.
HPE's CFO said it directly: "We wanted to select an end-to-end process where we could truly transform, not just solve for a single pain point," which captures the difference between theater and impact better than any consultant could phrase it.
The 11% of organizations with agentic AI in production aren't automating individual tasks or functions. They're redesigning entire processes and rebuilding governance around how machines operate. Amazon deployed its millionth warehouse robot coordinated by DeepFleet AI, improving travel efficiency by 10% in operations that are already optimized relentlessly, which is a data point that matters because it shows real ROI in the hardest possible environment. BMW factories have vehicles driving themselves through kilometer-long production routes doing actual work rather than just demonstrating capability, which is the kind of deployment that only happens when the technology is proven and the process has been redesigned to support it.
By 2028, agentic AI is forecast to make 15% of daily business decisions autonomously, which only happens if organizations solve three structural problems that many of them haven't even begun to grapple with. First, legacy systems need API modernization so machines can operate them reliably, because most enterprise systems were designed for humans with workflows that assume someone will fill the gaps when data is inconsistent or processes hit unexpected edge cases. Second, data architecture needs to change fundamentally, because nearly half of organizations report their data isn't searchable or reusable by AI systems, which means you're trying to run autonomous agents on data you can't reliably access or trust. Third, governance frameworks need clarity on when agents act autonomously and when they escalate to humans, because without audit trails, escalation paths, and clear rules, you're building risk rather than capability.
The organizations addressing these three problems in 2026 will have working systems by 2027 and competitive advantage by 2028. The ones that don't will join the 40% of failed projects, which is where the real sorting happens in this cycle.
DNA Data Storage: When Cold Archives Get Molecular
Most people hear "DNA storage" and file it under science fiction because it sounds like something out of a lab rather than infrastructure that companies will actually deploy, but the reality is more practical and more urgent than that sounds. The world is producing data faster than we can store it economically or sustainably, and data centers now consume city-block levels of power just to keep data that already exists accessible and cool, which means building more traditional storage at scale isn't sustainable anymore, especially for data that needs to last decades but gets accessed rarely or never.
DNA data storage solves a specific archival problem that's becoming urgent: storing enormous amounts of data at extreme density for extremely long periods without consuming power or degrading. One gram of DNA can store 215 million gigabytes, which is three to six orders of magnitude more compact than electronic memory hardware, and DNA lasts thousands of years without power or maintenance or any attention whatsoever, making it the densest known storage medium that exists.
What changed in 2025-2026 is that the technology moved from academic research where it was intellectually interesting to commercial deployment path where it's becoming economically viable, which is the transition that matters. Twist Bioscience miniaturized DNA synthesis using semiconductor-based platforms that produce more than one million DNA fragments on a chip the size of an iPhone, compared to traditional methods that generate only 96 in the same space, which compresses manufacturing cost in a way that changes the economics completely. DNA Script advanced enzymatic DNA synthesis, making it faster, safer, and easier to automate than chemical methods, which means it becomes an actual manufacturing process rather than a lab technique.
The reading speed of DNA using solid-state nanopores is expected to far exceed flash memory, and the economics are starting to work for cold storage where data gets written once and rarely accessed, which is exactly where the value is for archival applications rather than hot data that needs constant access.
Fraunhofer institutes are developing a microchip platform for mass data storage using synthetic DNA, which matters because it frames this as manufacturing and chip-scale process engineering rather than just biology, and that signal tells you commercialization is starting to be real. Biomemory is targeting mainstream DNA data storage for data centers beginning in 2030, and Iridia is building DNA storage chips designed to slot into data centers like solid-state drives, with backing from Seagate and Western Digital, which is the kind of signal from the actual storage industry that means this isn't speculation.
By 2028, DNA storage will be handling archival data for major tech companies, pharmaceutical research datasets, and government records, because the economics will work and the regulatory frameworks around data storage will adapt to support new technologies. The technology that could store all human knowledge in a shoebox is moving from lab to actual infrastructure deployment.
Post-Quantum Cryptography: The Migration Nobody Wants But Everyone Needs
Quantum computing creates a specific threat called "harvest now, decrypt later" that most people don't think about until they really need to, but the threat is concrete: adversaries can collect encrypted data today and store it until quantum computers become powerful enough to break current encryption, which might be years away but the data they're collecting now might need to stay secret for decades. For anything that must stay secret for years or decades like financial records, health data, state secrets, and intellectual property, this isn't theoretical risk that lives in security blogs, it's operational reality that finance teams and lawyers care about because the liability is real.
The shift that changed everything is that post-quantum cryptography moved from research topic that academics discuss to migration program that organizations are actually budgeting for and executing. In August 2024, NIST finalized its first three post-quantum cryptography standards, including ML-KEM for key encapsulation, ML-DSA for digital signatures, and SLH-DSA for stateless hash-based signatures, which turns PQC from "something to watch" into "something to implement" because the standards are fixed and organizations can stop studying it and start doing it.
What I expect in 2026-2028 is the unglamorous but essential infrastructure work that never makes headlines: inventories of where cryptography lives in systems, protocol upgrades that touch decades of code, certificate chain migrations that break things if you get them wrong, and painful compatibility bridges for long-lived devices that can't easily be updated because they're embedded in infrastructure that's been running for years. Government agencies have deadlines. The US Office of Management and Budget issued guidance requiring federal agencies to begin transitioning to post-quantum cryptography, which means the private sector companies that sell to them will need it too. Financial institutions are next in line, and then everyone else follows.
This won't make headlines or excite venture capitalists, but it's infrastructure work that has to happen and the organizations that treat this as compliance theater will get burned when quantum computers show up and render their encryption worthless. The organizations that do it properly will have cryptographic infrastructure that survives the quantum transition and doesn't need to be ripped out and replaced.
Fusion Energy: When Buyers Sign Contracts, Not Just Scientists
I've watched fusion promises for years with default skepticism because every decade brings new announcements that fusion is 20 years away, which is a pattern that becomes a joke after you see it enough times. What changed my perspective and made me actually believe fusion is happening wasn't another physics breakthrough that made headlines, it was buyers signing power purchase agreements, which is the kind of signal that matters because companies don't commit capital to binding contracts unless they believe the technology will deliver.
Helion Energy secured the world's first fusion power purchase agreement with Microsoft, targeting delivery in 2028, which isn't a research grant or a letter of intent that sounds good until the quarterly review, it's a commercial contract with penalties if power doesn't deliver on schedule. Microsoft needs that electricity for AI data centers that require power at scales that grids can't provide anymore, and they wouldn't sign a binding agreement without confidence in the technology and timeline because the reputational risk would be enormous.
Helion broke ground on its fusion power plant in Washington in July 2025 and received approvals for the next construction phase in October 2025, which shows this is moving from announcement to actual construction site deployment, and the company demonstrated fusion with its Polaris prototype in 2024, proving the physics works at scale in a way that passes engineering review. Commonwealth Fusion Systems is building SPARC, targeting net energy demonstration in 2026, which is a specific engineering milestone that's either going to happen or it won't, and the company uses high-temperature superconducting magnets that enable compact tokamak designs, which solves one of fusion's historical problems where you needed a reactor the size of a city to make the economics work.
The National Ignition Facility achieved ignition with Q=1.37 in 2022, proving fusion can produce more energy than it consumes, and multiple private companies are now targeting Q>10 for commercial viability because that's where the economics work for power generation.
Commonwealth secured power purchase agreements with utilities and announced plans for the world's first commercial fusion power plant in Virginia, with operations targeted for the early 2030s, which is the kind of statement you only make when you have investors, engineers, and customers who believe you can actually deliver. The US Department of Energy selected 11 developers for its Nuclear Reactor Pilot Program with goals to reach criticality for at least three reactors by July 2026, and President Trump issued executive orders to fast-track research and deployment, which shifts the regulatory environment from blocking innovation to enabling deployment rather than fighting it.
The first commercial fusion plants will come online between 2030-2035, and Commonwealth Fusion and other developers have announced specific timelines that their shareholders and customers will hold them accountable for meeting. The fusion energy sector could reach $40-80 billion by 2036, which is the kind of market size that attracts capital and keeps companies focused on execution rather than endless research.
Fusion's value proposition is straightforward and compelling if it works: unlimited fuel from seawater, no carbon emissions, no long-lived radioactive waste, and no meltdown risk, which makes it fundamentally different from everything else in the energy space. The technology works because ignition has been demonstrated. Engineering challenges are being solved by companies with billions in capital and binding customer commitments. First commercial contracts are signed and construction is starting.
By 2028, multiple fusion reactors will be demonstrating net energy at scales that matter for power generation, and the first commercial plants will be under construction rather than still in conceptual stage. This is no longer science fiction or research project, it's infrastructure deployment with committed capital and binding contracts from companies that matter.
Why These Four Forces Compound?
These trends don't exist in isolation or move independently from each other, they're interconnected in ways that multiply impact when they converge. Agentic AI systems will coordinate operations across digital twins of factories and supply chains, optimizing in real-time because the coordination problems become solvable when you have autonomous decision-making at scale. DNA storage will archive the massive datasets generated by AI systems and digital twin simulations, solving the data storage problem that would otherwise explode into unsustainable costs. Post-quantum cryptography will secure the communications infrastructure that everything else runs on, because none of this matters if adversaries can decrypt it. Fusion energy will provide the clean power AI data centers need to scale without carbon constraints or grid limitations.
The AI revolution creates exponential data growth that traditional storage can't handle sustainably. DNA storage solves the archival problem with solutions that last centuries. AI and automation scale operations to levels that legacy power grids can't support. Fusion provides the energy infrastructure without carbon penalties. Post-quantum cryptography protects it all from adversaries collecting data today to decrypt tomorrow. Each force depends on and amplifies the others, which is how you get compounding impact.
Organizations that understand these connections will move faster than those treating each as a separate initiative that can be solved independently, because they'll see how progress in one domain unlocks progress in others. The 2026-2028 window is when these forces transition from pilot projects and research demonstrations to scaled operations that matter for real businesses.
What Three Years of Predictions Taught Me?
Looking back at 2022 taught me three lessons that changed how I evaluate trends going forward, because these lessons explain more about prediction failure than any amount of hindsight analysis could.
Infrastructure takes longer than technology. Space tourism had working spacecraft in 2021 that could actually carry people to the edge of space and back, but Virgin Galactic still wasn't flying customers in 2026 because manufacturing capacity, regulatory frameworks, and operational reliability take years to build in sequence rather than parallel, and you can't skip steps just because the technology works. When I look at fusion now, I see working reactors that achieve net energy by 2026, but commercial deployment scaled to power grids doesn't happen until the 2030s, and that gap between technical proof and scaled infrastructure matters because it's where most predictions fail.
Speculation distorts every signal. The metaverse and NFT hype made it impossible to separate real value from narrative and marketing momentum, because prices peaked before utility existed and the market had to correct itself in ways that were painful but necessary to establish real foundation. Now I look for signed contracts, paying customers, and operational deployments rather than projections and pitches because those are the only signals that survive contact with reality.
Regulatory clarity matters as much as technology. DeFi had working technology in 2020 but it took until 2025 for regulatory frameworks to mature enough that institutions would participate without risking regulatory blowback, which means the technology was ready years before adoption became possible. Gene editing got FDA approval in 2023, but the regulatory pathway for rapid approvals only emerged in 2025, which created a two-year gap between approval and infrastructure. Post-quantum cryptography has NIST standards now, but migration will take years through organizations because it touches everything. Policy determines deployment speed as much as engineering does, sometimes more.
When I evaluate the four trends I'm highlighting for 2026-2028, I'm applying these lessons actively rather than just theoretically. Agentic AI has the technology but needs organizational redesign at every company that deploys it. DNA storage has working prototypes but needs manufacturing scale that doesn't exist yet. Post-quantum crypto has standards but needs migration execution across thousands of organizations. Fusion has net energy demonstrations and signed contracts but needs construction to complete on schedule and engineers to solve the engineering problems that haven't been solved yet. The gap between "it works" and "it's deployed at scale" is where predictions live or die.
The Separation Is Happening Now
For organizations deciding where to place bets, the message is direct and comes from watching what actually happened rather than what I thought would happen. Identify where you have real competitive stakes in these four forces, run disciplined pilots in 2026, and move to production by late 2027 because that's when the gap between leaders and laggards becomes permanent rather than recoverable.
- The 11% of companies with agentic AI in production will have structural advantages by 2028 that competitors can't replicate quickly because they'll have years of operational learning that's impossible to compress into a few months.
- The companies implementing DNA storage for cold archives now will have years of operational data when costs drop enough for broader adoption, which puts them ahead in the cost-efficiency curve.
- The organizations deploying post-quantum cryptography in 2026-2027 will be secure when quantum computers mature enough to matter, rather than facing crisis when that happens.
- The tech companies that secured fusion power contracts in 2024-2025 will have electricity available when competitors face grid constraints, which matters more than people think.
The gap between leaders and laggards doesn't narrow or stay stable, it widens exponentially because advantage compounds when you're right about technology deployment timelines and massive when you're wrong. The moves you make in the next 18 months determine which side of that gap you end up on, and there's no catching up from the lagging side once the separation is complete.
0 Comments