"Software is dead. AI agents will eat everything."
That sentiment, and the panic behind it, evaporated nearly half a trillion in software market value in weeks. Leaders and teams across the technology landscape are feeling it, viscerally. The panic is real, but the premise is wrong.
The prevailing narrative misunderstands the very nature of evolutionary innovation at the heart of software. It mistakes a great sorting for a great extinction. And it overlooks a clear pattern that has repeated across every major platform shift in the history of technology.
At ICONIQ, we have spent the last thirteen years learning, partnering early with generational companies across the full arc of software, from leading infrastructure platforms like Datadog, Snowflake, and Adyen to vertical pioneers like Procore and ServiceTitan. We have also led billions of dollars in investments in category-defining AI companies shaping the agentic era: Anthropic, ElevenLabs, Glean, Sierra, OpenEvidence, Legora, and others.
That is not a contradiction, nor a hedge. It reflects our conviction. We do not see an extinction but rather evolutionary innovation and a great sorting. Two distinct layers are emerging to create software’s agentic epoch. The companies that understand what comes next will build some of the most valuable enterprises of the coming decade.
The Paradox of Creation, The Great Sorting
Here is the paradox of this moment: AI will create exponentially more software. As a result, it will spectacularly raise the bar for what users demand from software.
There has always been a shortage of truly exceptional products.
Most enterprise software is tolerated, not beloved. Have you ever used ADP?
Enterprise software did something extraordinary: It standardized workflows, controls, and best practices directly into systems of record. Organizations gained efficiency, auditability, and global scale—but they also grew more alike. Companies adapted to the assumptions embedded in their software. The system defined the process.
The agentic era introduces a new possibility. AI allows systems to retain structural integrity while becoming more responsive to context and intent. Instead of requiring businesses to conform to static workflows, software can increasingly adapt to the nuances of how work actually happens. The foundation remains. The flexibility returns.
AI creation will only emphasize the importance of exceptional products. When anyone can produce a tool in an afternoon using Claude Code, the mediocre middle will be increasingly uninteresting. The products that embed AI to deliver extraordinary experiences while transforming to the operational nuance of each business will thrive.
This is the Great Sorting.
Some software darlings of the cloud era will not survive it. That is not pessimism; it is precedent. Businesses like Siebel did not fade because customer relationships stopped mattering. Salesforce rebuilt CRM for the cloud. Legacy monitoring gave way to cloud-native Datadog. On-premise data warehouses were pushed aside by the capabilities of Snowflake. In each case, the category endured, reimagined.
The very nature of software is one of evolutionary innovation. The greatest products are never finished. They constantly evolve at the edge of invention and in the direction of customer value. Velocity will increase dramatically with AI, but connecting these fast moving pieces into an amazing product is an art.
Those composing with the greatest taste and richest understanding of solving problems will unfairly participate in an uneven transition. Others may learn the lessons of Siebel.
Participation Has Always Been the Engine
The history of software is a chronicle of expanding participation and rising value.
The mainframe era concentrated computing power in the hands of a few specialists. Client-server architectures distributed it more broadly, creating formative operating systems and systems of record. The cloud expanded it dramatically, marked by rapid expansion in distribution and velocity for billions of users globally.
Each transition widened the aperture of access and participation. More users meant more engagement, more data, more workflows, and ultimately more value. Every platform shift came with orders-of-magnitude growth.
Agents represent the next horizon of participation.
The number of agents interacting with software will not scale linearly, but exponentially. Agents do not have the same onboarding time. They operate across systems and languages. Participation is about to explode in a way that makes even the cloud transition look modest by comparison.
For those building truly exceptional products, there are two clear paths of success in this new era.
One is to build valuable agents as a distinct new layer. We believe it will be difficult for core software systems to build best-in-class agents, but the very best will do so and integrate vertically.
The other path is equally promising: Embrace the agentic future to expand the reach and utility of amazing products through scaling dimensions never previously imagined. Fuel interdependent agents with structured data, context, and workflows. A mutualistic symbiosis will take hold.
Mutualistic Symbiosis and The Great Barriers
Two hundred million years ago, during the late Triassic period when the earth’s continents were one land mass, the Earth’s ocean was starved for nutrients. This stress created opportunity, giving rise to one of natural history’s most productive symbiotic relationships — coral and algae creating plentiful and expansive coral reefs to feed an exploding ocean economy.
Coral provides structure, protection, and essential nutrients. Algae convert sunlight into energy, feeding the coral and fueling the broader ecosystem. Neither thrives alone. Together, they built some of the most biodiverse and beautiful structures on Earth.
We see a parallel in software, two analogous layers forming, both distinct but deeply interdependent:
The Data Context Layer (“Coral”): Platforms with deeply rooted domain expertise, embedded workflow context, and hard-earned trust. These systems of record are the operational backbones, repositories of structured data and institutional knowledge that enterprises have built over years. They are the foundational coral that lines our sea bed, upon which all else is built.
The Agent Layer (“Algae”): AI systems that leverage the data context layer, converting the raw inputs to automate increasingly complex, long-form tasks. These algae free users to focus on judgment, creativity and strategy. They reason, plan and execute across multiple systems, enabling new modalities that dramatically expand software's impact.
Yes, there will be intrinsic friction and great ambitions to complicate things, but the relationship between these two distinct and interdependent layers can be beautifully symbiotic.
The platforms provide the foundation, the integrity, the context. Agents rely on structured data, embedded workflows, and operational intelligence within systems of record. The agents execute, connect and amplify. Together, they can generate outcomes neither could achieve alone.
It will be challenging for agents to build the data context layer. Deep domain expertise, institutional trust, and proprietary customer data. Those are earned over time. It will be hard for the data context layer to build agents, a completely new paradigm.
There will be some exceptional companies that will endeavor to span between layers. More broadly, we anticipate increasing interoperability, where value accrues to both. The most durable businesses will build from their strengths toward an agentic future.
A Coming Age of Reason
Capital markets leverage human extremes—fear and greed—and the volatility can be unsettling during drastic change. AI has intensified this.
In these moments, leaders need to exercise urgency with reason and act with conviction. The instinct to panic is human. The discipline to transform decisively and build exceptional products is what separates generational companies.
What we are witnessing is not the death of software. It is a Great Sorting. An expansion of what software can do, followed by a separation of who can do it at the highest level. Paradoxically, the benchmark for amazing products is only going up. The market is only getting bigger with agentic participation. And the companies that embrace the symbiosis between data context and agents—and build for interoperability—will be on the winning side of the Great Sorting.
The cloud gave rise to Salesforce, Workday, ServiceNow, CrowdStrike, Datadog, Snowflake, Databricks, and many others. The agentic era will produce its own leaders.
AI is not devouring software, it is reshaping and democratizing it—the Great Sorting has begun. We are entering a New Age of Reason. The companies that endure will be those that embrace the symbiosis to build exceptional experiences.
At ICONIQ, we are energized by the awe-inspiring coral reefs being created, and by the vast opportunity for so many across this expeditiously emerging oceanic economy.
Special thank you to Amit Agarwal, Rob Bernshteyn, Phil Saunders, Abi Noda, Alex Dhillon, David Pescovitz, Will Griffith, and Divesh Makan for your thoughts and suggestions.
Published:
February 17, 2026
.jpg)
.jpg)