Paraphrasing William Gibson, the future of AI is here, but it is nowhere close to evenly distributed yet. This core insight drives the current state of enterprise AI adoption, which is far more nuanced than simple narratives of rapid transformation or stagnant resistance.
Recent conversations in London crystallize this reality. The head of engineering at a large hedge fund described teams with fleets of AI agents in full production, and personally, all his code is now written by large language models (LLMs). Interestingly, junior hires are prohibited from using LLMs for code assistance—a deliberate strategy to build foundational skills. In stark contrast, a data engineer at a major retail bank reported the opposite: no agents and sparse LLM usage, with their division moving at a glacial pace. This divergence is not about one company “getting” AI while another fails; it reveals that even within the same organization, adoption curves vary wildly between teams.
This pattern is confirmed by data. McKinsey’s research indicates that 88% of organizations use AI in at least one business function, but only about one-third have begun scaling AI programs. For agentic AI, 23% report scaling somewhere in the enterprise, while 39% still experiment. In no function do more than 10% of respondents say they are scaling agents. Broad usage, therefore, does not equate to deep institutional change. There remains ample time to figure out AI without being left behind.
Cue the engineering boom
Common assumptions that “finance is cautious” or “regulated industries are behind” are oversimplifications. Some financial firms move aggressively, some do not, and some teams within the same firm do both simultaneously. Deloitte’s 2026 enterprise AI research reinforces this: only 25% of respondents have moved 40% or more of AI pilots into production. Just 34% claim to use AI to deeply transform their businesses—a number likely aspirational—while 37% still apply AI at a surface level with little change to core processes. This is not a tidal wave but a messy, uneven organizational test.
This unevenness directly challenges fears that AI will wipe out software jobs. The interesting dynamic is not that AI makes software cheaper to produce, but what companies do with that lower cost. Box CEO Aaron Levie invokes Jevons paradox: when a capability becomes cheaper and easier to consume, demand for it often rises. Cloud computing did not reduce compute needs; it spurred more applications. AI-assisted coding appears to be doing the same for software.
Engineering job data supports this. Lenny Rachitsky highlighted that engineering openings are at their highest levels in over three years. TrueUp data shows 67,665 open engineering jobs as of March 2026, a 78.2% increase from the recent low. Crucially, this growth is not only at the top: 44.6% of posted roles are entry- and mid-level, versus 38.3% senior and 13.8% senior-plus. AI is not eliminating junior roles; companies still want many engineers, even as AI tools proliferate.
Stack Overflow’s 2025 survey found 84% of developers use or plan to use AI tools, and over half of professionals use them daily. McKinsey reports that high-performing AI-driven software organizations see 16–30% improvements in productivity, customer experience, and time to market, along with 31–45% improvements in software quality. However, these gains come not from sprinkling copilots over unchanged processes, but from reworking roles, workflows, and the full product development system—a far harder organizational challenge than buying licenses.
Software engineering is alive and well
The hedge fund leader represents an early glimpse where engineering shifts from hand-authoring code to specifying, reviewing, steering, and orchestrating systems that generate code. But the retail bank division is not irrationally lagging; in heavily regulated environments, governance is the hard part. Deloitte reports that only 21% of companies have a mature governance model for autonomous agents (likely overestimating), while 73% cite data privacy and security as top risks, and 46% cite governance capabilities. This is not bureaucracy for its own sake; plugging non-deterministic systems into deterministic, compliance-heavy environments is messy.
Caution, however, is not free. Every quarter spent in pilot mode is a quarter where more aggressive peers build operational muscle. OpenAI’s enterprise usage data shows how uneven that muscle-building is: frontier workers (95th percentile) send six times more messages than the median worker; frontier firms send twice as many messages per seat. The primary constraints are no longer model performance or tools, but organizational readiness and implementation.
The real divide is between teams that have learned to integrate AI into repeatable work and teams that still treat it as a promising but dangerous sideshow. The distinction of task versus job is critical. Writing boilerplate code is a task; engineering is a job. Jobs bundle judgment, trade-offs, accountability, architecture, security, integration, testing, and the ugly reality of operating systems in the real world. AI automates more tasks but has not eliminated the need for such jobs, especially where bad software decisions carry operational or regulatory consequences. McKinsey’s broader survey reinforces that high performers redesign workflows and treat AI as a catalyst for innovation and growth, not just efficiency.
AI is not plodding toward a uniform future where software engineers fade away. Instead, it splits enterprises into fast-learning and slow-learning teams, rewarding those that redesign work, govern risk, and turn lower software costs into more software, not less. Code may be getting cheaper, but the ability to decide what to build, how to fit it together, and how to keep it from breaking the business keeps increasing in value. That is not the death of software engineering; it is the repricing of it, and every company and every team pays different prices.
Source:InfoWorld News