OpenAI President Greg Brockman’s claim that AI now handles 80% of coding tasks signals a profound shift in the software development labor market, moving from machine assistance to workflow automation.
Back
Back
OpenAI president says AI now writes 80% of code for engineers
(P1 - Lede) OpenAI President Greg Brockman said that artificial intelligence tools have advanced from being a minor support function to generating as much as 80% of code for software engineers. The statement points to massive productivity gains and reinforces the investment thesis for AI-centric companies like Microsoft, Google, and Nvidia, while signaling a fundamental restructuring of the technology labor market.
(P2 - Authority) "AI helps engineers do their jobs more effectively rather than replacing them," the Boston Consulting Group concluded in a recent report. This view is echoed by leaders like ServiceNow CEO Bill McDermott, who has committed to reskilling employees whose roles are impacted by agentic AI, transitioning them into management or other internal positions.
(P3 - Details) The impact is already visible in labor data, though the signals are conflicting. A Stanford study found a 16% decline in early-career employment for AI-exposed jobs since late 2022, and software development postings on the job site Indeed have fallen 53%. Yet, BCG found that overall software engineering headcount has still grown, albeit at a much slower 2% annual rate since ChatGPT's public release. Starting salaries for computer science majors are even expected to increase by almost 7% year-over-year, according to the National Association of Colleges and Employers.
(P4 - Nut Graf) The primary effect is not mass layoffs but a hiring slowdown that economists label a "big freeze." Companies are achieving higher output with their existing workforce, reducing the need for new hires and constricting the pipeline for entry-level talent. For investors, this trend could boost profit margins at large tech firms but also signals long-term risks to talent development and innovation if the entry-level job market continues to narrow.
The Shift to Agentic AI
The acceleration in coding capability stems from the move beyond simple generative AI, which handles discrete tasks like drafting text, to more advanced agentic AI. These systems can tackle broader objectives by breaking work into sub-tasks, moving across systems, and revising their approach with limited human input. The focus is shifting from task automation to complete workflow automation.
Major financial institutions are at the forefront of this adoption. JPMorgan Chase, with a technology budget of $19.8 billion, is deploying agentic systems in software engineering to give its developers more context to handle complex tasks. Lori Beer, the bank's global CIO, confirmed that senior engineers now spend more time creating specifications and reviewing AI-generated code, rather than writing it from scratch. The bank has already onboarded 200,000 employees onto an internal LLM suite to build their own AI assistants.
This pattern is consistent across sectors. Salesforce cut roughly 4,000 customer-service roles after AI agents began handling about half of customer interactions, and IBM eliminated 200 HR positions after its "AskHR" system automated routine employee inquiries. These are not broad cuts, but surgical reductions in workflows now managed end-to-end by AI.
A Hiring Freeze, Not a Firing Wave
While headlines often focus on job elimination, the more immediate impact is a sharp slowdown in hiring, which has dropped to levels last seen in 2010 when unemployment was near 10%. Companies are not firing existing staff but are quietly freezing the replacement of workers who leave. A recent McKinsey survey found that while 43% of companies expect AI to have no effect on workforce size, 32% expect to decrease their employee base by at least 3% within a year—a reduction that can largely be met through natural attrition.
This creates a paradox for the labor market. Unemployment remains near historic lows at around 4%, but job market confidence has deteriorated. The share of U.S. workers who believe it is a good time to find a quality job has fallen from 70% in 2022 to just 28% recently. College graduates are now more pessimistic than those without degrees, a reversal of historical trends. The result is a growing sense of stagnation, where fewer entry-level pathways exist to gain experience and advance.
Open vs. Closed AI Strategies
As companies integrate AI into their development cycles, a strategic divide is emerging between open and closed approaches. Canonical, the company behind Ubuntu Linux, is integrating AI with a clear preference for open-weight models and on-device inference. This strategy, outlined by VP of Engineering Jon Seager, prioritizes user control and privacy, allowing developers to choose which AI tools to use and run them locally.
This contrasts sharply with Microsoft's strategy, which anchors its Copilot services to its proprietary Azure cloud. While powerful, this approach creates vendor lock-in and centralizes data processing. For investors, this divergence presents a choice: the integrated, high-margin ecosystem of Microsoft versus the flexible, potentially lower-cost open-source model championed by companies like Canonical. The success of these competing philosophies will shape the future of software development and the multi-billion dollar market for AI tools.
This article is for informational purposes only and does not constitute investment advice.
[1] OpenAI's president says AI has gone from writing 20% to '80% of your code'[2] Silicon Valley Is Bracing for a Permanent Underclass[3] AI won’t kill your job — it will kill the path to your first one[4] How JPMorgan’s CIO is reshaping work at the bank with a $19.8 billion annual tech and AI budget[5] Canonical's approach to AI is refreshingly thoughtful - Microsoft should take note
