March quarter revenues reached $21.6 billion, a 27% year-on-year growth — the highest rate in five years — and net income jumped dramatically to $521 million. The company's Hong Kong shares surged nearly 20% in a single session, becoming the biggest percentage gainer on the Hang Seng index that day. But the number that best explains the market's reaction is not in the margins or PC volumes: it's the fact that AI-related revenues grew 84% in the quarter and accounted for 38% of the group's total revenues.
The conversation about artificial intelligence in large enterprises follows a comfortable script: evaluating platforms, approving budgets, designing pilots. Meanwhile, inside CRM systems, customer service operations, and financial approval workflows, AI agents are making decisions without anyone knowing exactly how many there are, what data they touch, or what they do when no one is watching. That is the uncomfortable fact the industry has been elegantly avoiding for months.
Since late 2022, Asian markets have undergone a silent but profound reconfiguration. The emergence of generative artificial intelligence not only transformed the narrative of global markets, but reordered the specific weight of regional indices around a handful of names. Three companies — Taiwan Semiconductor Manufacturing Company, Samsung Electronics, and SK Hynix — came to explain more than half of the returns of the FTSE Asia ex-Japan index.
On May 13, 2026, Anthropic launched Claude for Small Businesses, a version of its AI assistant connected directly to the operational tools of small businesses: email, calendar, and — this is what's new — accounting software. The concrete promise is that Claude can perform reconciliations, generate profit and loss statements, and categorize transactions without the owner having to touch a spreadsheet. But the reaction from the specialized market was not one of unqualified enthusiasm: it was a cautious welcome, with a warning that has been echoing through this sector for some time.
There's a narrative that has dominated boardroom conversations and venture capital funds for two years: artificial intelligence will devour enterprise software the same way software devoured analog business models. It's a powerful image. And like every powerful image that circulates without friction, it deserves pressure before it dictates investment decisions with real consequences.
There's a scene that repeats itself in almost every mid-sized company I know. The technology team presents an artificial intelligence pilot. The initial numbers look promising. The board approves the investment. And six months later, the pilot is still just a pilot.
In a week in May 2026, enterprise AI infrastructure crossed a boundary that audit, compliance, and insurance frameworks had not yet drawn. On May 7, AWS previewed Amazon Bedrock AgentCore Payments, a system built with Coinbase and Stripe that allows artificial intelligence agents to make autonomous payments during execution. Two announcements in seven days, from two of the largest technology infrastructure platforms on the planet, describe the same behavior: an agent that decides to spend money on its own.
More than $100 million for a daily tech show that generates approximately $5 million in annual revenue. That is a valuation multiple of over 20x on sales for a media asset, in a sector where typical multiples rarely exceed 3x or 4x revenue. This is not a miscalculation. It is a strategic statement.
There is a pattern that repeats with enough consistency to take seriously: technologies do not concentrate where they are seen, but where they are supported. Social networks concentrated on distribution, not content. The cloud concentrated on infrastructure, not applications. Artificial intelligence is following the same geometry, but the control point is one level deeper than in any previous cycle.
The dominant narrative about artificial intelligence and business has a structural bias that is rarely named: it is built almost exclusively around companies with more than 500 employees. Not because large corporations are more interesting, but because for technology vendors they represent more predictable contracts, relatively shorter sales cycles, and recurring revenue streams that justify sales and marketing spend. The logic is understandable from the seller's economics. The problem is that this logic has distorted the reading of where real work happens in the economy.
There is a silent pattern that economic history has repeated at least twice before the era of artificial intelligence. First with industrial electrification, then with personal computers. In both cases, the technology arrived decades before its impact appeared in productivity statistics.
Quantum computing has spent more than a decade promising to reshape medicine, materials, and artificial intelligence. During that time, most capital flowed toward the superconducting circuits of IBM and Google, platforms requiring cooling to temperatures near absolute zero, costly infrastructure, and constant calibration. But beneath that dominant narrative, a different bet was taking shape: using neutral atoms as qubits, trapping them with lasers, operating them at room temperature, and scaling them into arrays of hundreds or thousands of units.
There comes a moment in the life of any productivity platform when doing one thing well is no longer enough. Notion has reached that point. The company—known for years as the place where teams store notes, wikis, and databases—has just announced a deep reconfiguration of its architecture: a set of capabilities that, taken together, transform the workspace into an environment where artificial intelligence agents can operate, receive instructions, execute code, and sync external data in continuous real time.
One night in late 2024, Denis Shilov was watching a crime thriller when an idea struck him. He wrote a prompt that caused any AI model to ignore its own safety filters. What Shilov concluded from that episode was not that he had found a bug, but that no company had a post-deployment control layer over what their AI models were doing once users started interacting with them.
There is a fact that should make any executive who has approved an artificial intelligence budget in the last two years uncomfortable: the United States, the country that builds the world's most powerful models, ranks 24th in global AI adoption. Its rate is 28.3%. The problem is not technological. It never was.
There is a pattern that repeats itself every time a technology stops being an experiment and becomes production infrastructure. It happened with relational databases, with cloud services, with microservices. And now it is happening with large language models.
The conversation around enterprise artificial intelligence security tends to converge on the same points: poorly trained models, hallucinations, algorithmic bias. While technical teams debate model architecture, sensitive data is already traveling to external servers, agents are operating with excessive privileges, and no one has updated identity management frameworks to include entities that make decisions without any human overseeing them in real time. The gap is not technical in origin. It is behavioral and organizational.
When a company generates $97 billion in free cash flow in a single fiscal year, the question is not whether it can invest. The question is what power architecture it builds with that money and who gets trapped inside it. Nvidia crossed $40 billion in capital commitments in the first five months of 2026, including a $30 billion bet on OpenAI.
Sarvam AI, Ebix's X Pay and AuthBridge's AuthLead address different business problems. Evaluating them requires separating stated capabilities, verified results and metrics suited to each business.
There is a belief that runs through the corridors of almost every organization that has invested in artificial intelligence over the last eight years. The belief that the problem is always about quantity. More data. More tokens. More coverage. More stored history.
When SAP shells out $1.16 billion for an 18-month-old German startup, it's not buying technology. It's buying time. And when Anthropic and OpenAI announce, in the same week, their own structures to bring AI to large enterprises, what emerges is not a race for the best model — it's a race for who controls the layer where business decisions get automated.
There is a precise moment in the cycle of any business model where the collective narrative stops describing reality and starts producing it. The SaaS sector reached that moment more than a year ago, and the industry is still processing what it means. It is not the collapse that some anticipated with the term 'SaaS-pocalypse', but neither is it a frictionless return to 2021-era growth.
The image that best describes the state of artificial intelligence in businesses during 2025 is not one of a technology that failed. It is one of a technology that was used without real commitment. According to an MIT report published that year, 95% of generative AI pilots never reached production with measurable impact.
Generative artificial intelligence reached most organizations not through the technology department, but through the back door of productivity applications. Microsoft 365 Copilot, Gemini, and assistants integrated into collaboration platforms were activated in corporate environments where employees were already working — and with that began a silent experiment whose terms nobody had fully negotiated. The problem is not with the language models. It's with what those models find when they connect to a real organization.