Imagine 2026: Part Two
God Translation Bureau is a compilation team under 36Kr, focusing on fields such as technology, business, workplace, and life, and mainly introducing new technologies, new ideas, and new trends from abroad.
Editor's note: AI is no longer just a chat box. By 2026, it will penetrate the underlying layers of factories and banks, completely "overhauling" the physical world along with business models. This is the second article in a series, and it is a compilation.
American Dynamism Team
David Ulevitch: Building Native AI Industry Infrastructure
The United States is reshaping the economic sectors that create real strength. Energy, manufacturing, logistics, and infrastructure are back in the spotlight, and the most important shift is the rise of truly native AI and software - first industry infrastructure. These companies start with simulation, automated design, and AI - driven operations. They are not modernizing the past but building the future.
This opens up huge opportunities in fields such as advanced energy systems, heavy robot industries, next - generation mining, and biological and enzymatic processes (producing precursor chemicals that various industries rely on). AI can design cleaner reactors, optimize mining, engineer better enzymes, and coordinate autonomous machine clusters with insights that traditional operators cannot match.
The same transformation is reshaping the world outside factories. Autonomous sensors, drones, and modern AI models can now achieve continuous visibility of ports, railways, power lines, pipelines, military bases, data centers, and other aspects that were once too large to manage comprehensively.
The real world needs new software. The founders who build this software will shape the prosperity of the United States in the next century. If you are such a person, talk to us.
Erin Price - Wright: The Revival of American Factories
The century - long prosperity of the United States is built on industrial strength, but as we all know, we have lost most of that strength - partly due to offshoring and partly due to the society's deliberate avoidance of construction. But the rusty wheels are starting to turn again, and we are witnessing the rebirth of American factories centered around software and AI.
In 2026, I think we will see companies adopt a "factory mindset" to address the challenges in energy, mining, construction, and manufacturing. This means deploying AI, automation technologies, and skilled workers in a modular way, making complex and customized processes run like an assembly line. Specifically:
Handle complex regulations and permits quickly and repeatedly.
Accelerate the design cycle and start with "design for manufacturing" from the beginning.
Better manage large - scale project coordination.
Deploy automation technologies to accelerate tasks that are difficult or dangerous for humans.
By applying the techniques developed by Henry Ford a century ago, planning for scale and repeatability at the start of a project, and integrating the latest advances in AI, we will soon be able to mass - produce nuclear reactors, build housing to meet national needs, construct data centers at an astonishing speed, and enter a new golden age of industrial strength. In the words of Elon Musk: "The factory itself is the product."
Zabie Elmgren: The Next Wave of Observability Will Be Physical, Not Digital
In the past decade, software observability has changed the way we monitor digital systems, making codebases and servers transparent through logs, metrics, and tracing. The same revolution is happening in the physical world.
With over a billion connected cameras and sensors deployed in cities across the United States, physical observability - that is, having real - time knowledge of what is happening in cities, power grids, and other infrastructure - is becoming both urgent and feasible. This new sensing layer will also open up new frontiers for robotics and automation technologies. Machines will rely on a common architecture to make the physical world as observable as code.
Of course, this transformation also brings real risks: the tools that can detect forest fires or prevent workplace accidents may also fuel utopian nightmares. The winners in the next wave will be those who can earn public trust, building systems that protect privacy, are interoperable, and are native to AI, enhancing social clarity without depriving people of their freedom. Those who can build this trusted architecture will define observability in the next decade.
Ryan McEntush: The Electro - Industrial Stack Will Move the World
The next industrial revolution will not only happen in factories but also inside the machines that drive them.
Software has changed the way we think, design, and communicate. Now, it is changing the way we move, build, and produce. Advances in electrification, materials, and AI are converging, bringing real software control into the physical world. Machines are starting to sense, learn, and act autonomously.
This is the rise of the electro - industrial stack - the integrated technology that powers electric vehicles, drones, data centers, and modern manufacturing. It connects the "atoms" of the world with the "bits" that command it: minerals are refined into components, energy is stored in batteries, electricity is guided by power electronics, and power is output through precision motors, all coordinated by software. It is the invisible cornerstone behind every physical automation breakthrough; it is the fundamental difference between "software that just hails a taxi" and "software that takes over the steering wheel."
However, the ability to build this technology stack, from refining key materials to manufacturing advanced chips, is being lost. If the United States wants to lead the next industrial era, it must manufacture the hardware that supports it. The countries that master the electro - industrial stack will define the future of industrial and military technologies.
Software has devoured the world. Now, it will move the world.
Oliver Hsu: Autonomous Laboratories Accelerate Scientific Discovery
As model capabilities progress across various modalities and robotic operation capabilities continue to improve, teams will accelerate the pursuit of autonomous scientific discovery. These parallel technologies will give rise to autonomous laboratories that can complete scientific discovery in a closed - loop - from hypothesis development to experimental design and execution, then to reasoning, analyzing results, and iterating future research directions. The teams building these laboratories will be interdisciplinary, integrating expertise in AI, robotics, physical and life sciences, manufacturing, operations, etc., and conducting continuous exploratory experiments in various fields through "lights - out" laboratories.
Will Bitsky: "Data Expeditions" in Key Industries
In 2025, the zeitgeist of AI was defined by computing power constraints and data center construction. By 2026, it will be defined by data constraints and the new frontiers of data expeditions - our key industries.
Our key industries are still a source of latent and unstructured data. Every truck dispatch, meter reading, maintenance operation, production run, assembly, and test firing is material for model training. However, data collection, annotation, and model training are not yet common terms in the industrial world.
There is no lack of demand for this type of data. Companies like Scale and Mercor, as well as AI research laboratories, are greedily collecting process data (not just "what" was done, but also "how"). They are paying a high price for every unit of painstakingly collected data.
Industrial companies with existing physical infrastructure and labor have a comparative advantage in data collection and will start to leverage this. Their operations generate a large amount of data, which can be collected at near - zero marginal cost for training their own models or licensing to third parties.
We can expect startups to offer help. Startups will provide a coordination stack: software tools for collection, annotation, and licensing; sensor hardware and SDKs; reinforcement learning (RL) environments and training pipelines; and ultimately, their own intelligent machines.
Applications
David Haber: AI Strengthens Business Models
The best AI startups are not just automating tasks; they are amplifying the economic benefits for customers. Take contingency - fee legal services as an example. Law firms only make money when they win cases. Companies like Eve use proprietary litigation outcome data to predict case success rates, helping law firms select better cases, serve more clients, and increase the frequency of wins.
AI enhances the business model itself. It drives more revenue, not just reduces costs. By 2026, we will see this logic extended to all industries as AI systems further align with customer incentives and create compound advantages that traditional software cannot match.
Anish Acharya: ChatGPT Becomes an AI App Store
The cycle of consumer products requires three things to work: new technologies, new consumer behaviors, and new distribution channels.
Until recently, the AI wave has met the first two conditions, but there has been no new native distribution channel. Most product growth has relied on existing networks like X (formerly Twitter) or word - of - mouth.
However, with the recent release of the OpenAI Apps SDK, Apple's support for mini - programs, and ChatGPT's launch of group chat functionality, consumer - side developers can now directly reach ChatGPT's 900 million users and leverage new mini - program networks like Wabi for growth. As the last piece of the puzzle in the consumer product cycle, this new distribution channel will kick off a once - in - a - decade gold rush in the consumer technology field in 2026. You bear the risk of ignoring it.
Olivia Moore: Voice Agents Shine
In the past 18 months, the concept of AI voice agents managing real - world interactions for businesses has gone from science fiction to reality. From small and medium - sized enterprises to large corporations, thousands of companies are using voice AI for appointment scheduling, bookings, surveys, check - ins, etc. These agents save companies costs, generate additional revenue, and free human employees to engage in higher - value - and more interesting - tasks.
But since this field is still in its infancy, many companies are still in the "voice - as - an - entry - point" stage, offering only one or two types of calls as a single - point solution. I'm very excited to see voice agents expand to handle entire workflows (possibly multi - modal) and even manage the entire customer relationship cycle.
This may involve agents deeply integrated with business systems and given the freedom to handle more complex types of interactions. With the continuous improvement of underlying models - and the fact that agents can now call tools and run across systems - there is no reason why every company shouldn't have a "voice - first" AI product to run and optimize key parts of its business.
Marc Andrusko: The Era of Prompt - Free and Proactive Applications Arrives
2026 marks the end of the "prompt box" era for mainstream users. The next wave of AI applications will have no visible prompting process - they will observe your actions and proactively intervene, offering actions for you to review. Before you even speak, your IDE suggests a refactoring; when you finish a call, your CRM drafts a follow - up email; when you're working, your design tool generates variations. The chat interface is just the "training wheels." Now, AI has become an invisible scaffold woven into every workflow, activated by intent rather than commands.
Angela Strange: AI Will Eventually Upgrade Banking and Insurance Infrastructure
Many banks and insurance companies have integrated AI such as document ingestion and voice agents on top of their traditional systems, but AI cannot truly change this industry unless we rebuild the underlying infrastructure that supports financial services.
By 2026, the risk of not modernizing to fully leverage AI will exceed the risk of failure. We will see large financial institutions terminate their old vendor contracts and implement newer native AI alternatives. These companies are not bound by past category boundaries and are platforms that can centralize, standardize, and enrich data from traditional systems and external sources.
What will be the result?
Workflows can be greatly simplified and parallelized. There is no need to jump between different systems and screens. Imagine: you can see and process hundreds of tasks that need to be completed in the loan origination system (LOS) in parallel, and agents can even complete the more boring tasks.
The categories we know will merge, creating larger categories. For example, customer identity verification (KYC) and transaction monitoring data from the onboarding process can now co - exist in a risk platform.
The new winners in these categories will be 10 times the size of the old established companies: because the market is much larger, and software is "devouring" the workforce.
The future of financial services lies not in applying AI to old systems but in building a new operating system based on AI.
Joe Schmidt: The Forward - Deployed Model Will Bring AI to the 99%
AI is the most exciting technological breakthrough we'll see in our lifetime. However, so far, most of the benefits brought by new startups have gone to the 1% of companies in Silicon Valley - either literally in the Bay Area or part of that extended network. This makes sense: startup founders want to sell their products to companies they know and can easily reach, whether by driving to the office or through board - level VC connections.
In 2026, this situation will reverse. Companies will realize that the vast majority of AI opportunities exist outside Silicon Valley. We will see new founders use the forward - deployed model to explore more opportunities hidden in large, traditional vertical industries. The opportunities will be huge in traditional consulting and service industries (such as system integration and implementation companies) and slower - paced industries like manufacturing.
Seema Amble: AI Creates a New Orchestration Layer and New Roles in Fortune 500 Companies
By 2026, enterprises will further shift from isolated AI tools to multi - agent systems that need to operate like coordinated digital teams. As agents start to manage complex, interdependent workflows (such as joint planning, analysis, and execution), organizations need to rethink the structure of work and how context flows between systems. We've already seen companies like AskLio and HappyRobot doing this, deploying agents across the entire process rather than for single tasks.
Fortune 500 companies will feel this transformation most acutely: they have the deepest siloed data, institutional knowledge, and operational complexity, most of which exists in people's minds. Transforming this background into a shared base for autonomous workers will enable faster decision - making, shorter cycles, and end - to - end processes that no longer rely on constant human micromanagement.
This transformation will also force leaders to re - imagine roles and software. New functions will emerge, such as AI workflow designers, agent supervisors, and governance leaders responsible for orchestrating and auditing the coordination of digital employee clusters. On top of today's systems of record, enterprises will need systems of coordination: a new layer for managing multi - agent interactions, determining context, and ensuring the reliability of autonomous workflows. Humans will focus on handling extreme situations and the most complex cases. The rise of multi - agent systems is not just another step in automation; it represents a reconstruction of how enterprises operate, make decisions, and ultimately create value.
Bryan Kim: Consumer - Grade AI Shifts from "Help Me" to "Understand Me"
2026 marks the shift of major consumer - grade AI products from productivity to a sense of connection. AI is no longer just helping you work; it is helping you see yourself more clearly and build stronger relationships.
Frankly, this is difficult. Many social AI products have been launched but failed. But thanks to multi - modal context windows and lower inference costs, AI products can now learn from all - around details of your life, not just what you tell the chatbot. Imagine an album showing real emotional moments, one - on - one private messages and group chat modes changing according to the chat partner, and daily habits adjusting under stress.
Once these products are truly implemented, they will become part of our daily lives. Usually, "understand me" products have better native retention mechanisms than "help me" products. "Help me" products are monetized through users' high willingness to pay for specific tasks and optimize subscription retention. "Understand me" products are monetized through daily engagement and continuous connection: although the willingness to pay is lower, the usage model is more retention - friendly.
People have always been trading data for value: the question is whether what they get in return is worth it. The answer will soon become clear.
Kimberly Tan: New Model Primitives Enable Previously Impossible Companies
In 2026, we will see the rise of companies that could not have existed before model breakthroughs in areas such as inference, multi - modality, and computer use. So far, many industries (such as law or customer support) have used improved inference capabilities to enhance existing products. But only now are we starting to see companies whose core product capabilities are fundamentally driven by these new model primitives.
Advances in inference capabilities can unlock new abilities to evaluate complex financial claims or handle esoteric academic and analytical research (such as adjudicating bill disputes). Multi - modal models make it possible to extract latent video data from industries rooted in the physical world (such as camera data from manufacturing sites). And "computer use" capabilities enable the automation of large - scale industries where value has traditionally been trapped behind desktop software, poor APIs, and fragmented workflows.
James da Costa: AI Startups Serving AI Startups Achieve Scale
We are in an unprecedented entrepreneurial moment driven by the current AI product cycle. But unlike previous cycles, traditional giants are not taking it lightly; they are also embracing AI. So how can startups win?
One of the most powerful and underestimated ways for startups to win channels from traditional giants is to serve them at the very beginning of their establishment: that is, greenfield companies (brand - new enterprises). If you attract new companies when they are founded and grow with them, when your customers become large companies, you will also become a large company. Stripe, Deel, Mercury, Ramp, etc. have all followed this strategy