Short answer: A reported 2024 recording quoted Matt Garman, then CEO of Amazon Web Services, saying it was “possible that most developers are not coding” roughly 24 months later. He did not say that all human programmers would disappear. By August 2026, AI coding tools had expanded substantially, but the available evidence does not prove that most professional developers had stopped coding—or that software engineers had become obsolete.
What the leaked audio reportedly said
The story began with a Futurism report published on August 22, 2024, which attributed excerpts to leaked audio from an AWS “fireside chat.” The available reporting, in turn, attributed the audio to Business Insider. No publicly accessible transcript or recording is established by the supplied source material, so the remarks should be treated as reported excerpts rather than independently authenticated quotations.
As an Amazon Associate I earn from qualifying purchases.
The most important reported line was conditional:
“If you go forward 24 months from now, or some amount of time—I can’t exactly predict where it is—it’s possible that most developers are not coding.”
Recommended: PC Feels Slow? A Free Scan Shows What's Dragging Windows Down →Recommended: Crashes or Glitches? A Free Driver Scan Usually Finds the Culprit →Recommended: Fix Windows Errors and Clear Junk Files in Minutes - Free Scan →Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.
Garman also reportedly said that “coding is just kind of like the language that we talk to computers,” and that developers would increasingly focus on customers, innovation, and deciding what to build rather than simply writing code.
Because the report appeared in August 2024, the 24-month horizon pointed approximately to August 2026. “Possible” matters here. This was a forecast about how developers might spend their working time—not a categorical announcement that programmers would lose their jobs.
Why the wording caused alarm
Garman was speaking as the head of AWS, Amazon’s cloud-computing division, shortly after becoming its CEO. The comments also appeared amid concern about Amazon workforce reductions and the rapid adoption of generative AI. That context made readers understandably ask whether the remarks represented a plan to eliminate software-engineering roles.
The reporting does not establish that Garman announced a companywide program to remove developers, nor does it prove that AWS layoffs were caused by AI. A prediction about the future composition of software work is different from a formal employment policy.
For workers, however, the distinction between “fewer routine tasks” and “fewer employees” is important but not reassuring by itself. If one engineer can complete more work with AI, a company might use the gain to build more products, reduce costs, slow hiring, or cut headcount. The technology determines what is possible; management decisions determine how the savings or additional output affect jobs and pay.
AWS’s interpretation: less routine work, not no developers
AWS characterized the remarks as a vision for removing “undifferentiated heavy lifting” so builders could spend more time on innovative work. That interpretation treats AI as a productivity tool: the machine handles repetitive implementation while people handle product judgment, customer needs, architecture, and other higher-value decisions.
That framing is also commercially relevant. AWS sells AI development software, including Amazon Q Developer. Its business interest is to make AI-assisted development more capable and widely adopted. That does not make Garman’s forecast wrong, but it is a reason to distinguish a vendor’s vision from independently measured changes in employment or developer productivity.
What “not coding” could mean
The phrase is ambiguous. It could describe several different changes:
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →- AI-assisted implementation: A developer describes a feature, reviews generated code, and makes targeted changes instead of writing every line manually.
- Higher-level orchestration: A developer manages AI agents, repositories, APIs, cloud resources, tests, deployment pipelines, and permissions.
- Role specialization: Some developers spend more time on architecture, security, reliability, product discovery, or industry-specific knowledge.
- Less typing, not less responsibility: People remain accountable for whether software is correct, secure, maintainable, compliant, and safe to operate.
None of these meanings is equivalent to “human programmers are a thing of the past.” Writing source code is only one part of software engineering. Requirements, design, testing, deployment, incident response, maintenance, and communication can consume as much or more effort than initial implementation.
What AI coding tools can do in 2026
AI development tools have moved beyond simple autocomplete. AWS says Amazon Q Developer can help generate and modify code, write tests and documentation, troubleshoot problems, scan for vulnerabilities, refactor applications, and assist with upgrades and modernization.
AWS also announced in May 2025 that Amazon Q Developer could work with GitHub issues to implement features, create bug fixes, review pull requests, and help modernize Java applications. Agentic workflows may read and write files, run shell commands, and interact with AWS services. Those capabilities allow an AI system to complete multi-step tasks, but they do not demonstrate that human oversight is unnecessary.
GitHub makes a similar distinction in its Copilot documentation. It describes Copilot as an efficiency tool, not a replacement for developers or a system that fully automates software development. GitHub warns that generated suggestions may contain bugs, insecure patterns, hardcoded credentials, SQL injection, path-injection vulnerabilities, outdated APIs, or other undesirable code.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →That warning is not a minor technical footnote. An agent that can change a repository or run commands has useful power, but it also needs carefully limited permissions, logging, testing, review, and rollback procedures.
What still requires substantial human judgment
AI can produce plausible code even when it has misunderstood the real problem. The difficult work often begins before a line of code is written.
- Defining the requirement: Stakeholders may disagree about what a product should do, which customers matter, or what risks are acceptable.
- Choosing an architecture: Teams must balance cost, latency, reliability, scalability, compliance, and operational complexity.
- Understanding legacy systems: Important behavior may exist in undocumented services, data pipelines, contracts, or workarounds that are difficult to infer safely.
- Verifying security and privacy: Generated code can expose data, mishandle authentication, or introduce vulnerabilities that ordinary tests miss.
- Debugging across systems: Failures may involve networks, queues, databases, cloud permissions, third-party services, and deployment configuration at the same time.
- Operating the result: Someone must respond when the system fails at 2 a.m., customers lose data, or a release creates a regulatory problem.
- Maintaining software: Dependencies, regulations, business priorities, and infrastructure change long after the original code is generated.
These are not proof that AI will never automate parts of the work. They explain why reducing manual code entry is not the same as eliminating engineering judgment or accountability.
Did Garman’s two-year forecast come true?
As of the available evidence through August 18, 2026, the strongest defensible conclusion is that the forecast was directionally plausible but not proven in its strongest form.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
AI tools clearly expanded from code suggestions toward agentic implementation, testing, review, refactoring, and modernization. However, the supplied evidence does not establish that most professional developers had stopped coding by August 2026. It also does not establish that human programmers had become obsolete or that developer employment had broadly collapsed.
Several claims that are often blended together should be kept separate:
| Claim | What it would actually show |
|---|---|
| Developers use AI to generate more code | Adoption of an assistance tool |
| Developers write less routine code manually | A change in the composition of daily work |
| AI completes larger features | Greater automation of implementation tasks |
| Companies hire fewer junior developers | A labor-market change, potentially caused by several factors |
| Most developers no longer code | A broad workflow change requiring current surveys or company-level data |
| Human programmers disappear | A far stronger claim that the available evidence does not support |
Measuring the forecast properly would require more than counting generated lines of code. Useful measures include acceptance and rework rates, defects, security incidents, delivery time, maintenance costs, model and review costs, and whether the AI works on isolated functions or complex production systems.
The absence of that evidence does not prove the forecast failed. It means the supplied material cannot responsibly call it fulfilled.
What this means for developers and their finances
For individual workers, the practical risk is not necessarily that programming disappears overnight. It is that the market may place less value on repetitive implementation and more value on the ability to specify, verify, integrate, and own software.
Developers can prepare by:
- Learning to use AI tools while retaining the ability to read, debug, and explain code.
- Becoming strong at testing, observability, security, and threat modeling.
- Understanding how repositories, CI/CD systems, cloud services, APIs, and permissions fit together.
- Building system-design skills rather than relying on generated snippets without understanding their consequences.
- Developing expertise in a business or technical domain that requires context and judgment.
- Improving communication with customers, product managers, and nontechnical stakeholders.
- Tracking whether AI actually improves delivery and quality instead of assuming that more generated code means more value.
Junior developers may face a particular trade-off. AI can help them learn and complete small tasks, but outsourcing every implementation decision may remove opportunities to develop debugging, code-reading, and design skills. A useful rule is to use AI to explain, test, challenge, and accelerate work—not to surrender understanding of the result.
Amazon Q Developer versus GitHub Copilot
The commercial details are secondary to the labor question, but the products illustrate how the market is evolving.
AWS lists a perpetual free tier for Amazon Q Developer with 50 agentic requests per month, subject to its stated limits, and lists Q Developer Pro at $19 per user per month on the pricing page reviewed for this article. AWS also lists Java transformation allowances and additional usage terms. Prices and limits can change, so readers should verify the current AWS pricing page before purchasing.
Amazon Q is most naturally suited to teams already operating heavily in AWS, particularly those that need AWS-resource context, IAM controls, or Java modernization. GitHub Copilot is a more natural fit for teams centered on GitHub repositories, pull requests, Actions, and enterprise administration. Neither product removes the need for review, testing, security controls, or clear ownership.
AWS also says on a product page that support for Amazon Q Developer IDE plugins is scheduled to end on April 30, 2027, with users directed toward Kiro for similar capabilities. That product-transition detail should be checked directly with AWS before making a buying decision.
The bottom line on the “programmers are finished” claim
Matt Garman’s reported remarks were a conditional forecast that AI could make manual coding less central to most developers’ jobs within about two years. They were not a prediction that all human programmers would vanish.
By 2026, AI coding systems had become capable of generating and modifying code, running tests, reviewing changes, and handling parts of multi-step development workflows. But the available evidence does not show that most developers stopped coding or that engineering judgment became unnecessary. The more credible scenario is a shift in the job: less routine typing, more specification, review, architecture, security, testing, operations, and domain expertise.
Free tools Windows power users keep installed
One-click scans. No signup required.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
