AI Is Replacing Developers — But the Real Numbers Tell a More Complicated Story
📑 Table of Contents
🎯 The 30-Second Summary
AI is reshaping software development faster than any previous technology wave. Entry-level developer jobs are down ~20% since 2024. 75% of Google’s new code is AI-generated. 92,000+ tech workers laid off in Q1 2026, with ~48% of cuts directly attributed to AI. But 1.3 million new AI-related roles were created in the last two years. India’s AI engineering hiring is up 59.5% year-on-year. Agentic AI job postings grew 10,854%. The honest answer: AI is replacing the bottom rung of the career ladder — while aggressively building new rungs above it.
On the morning of April 15, 2026, roughly 1,000 people who worked at Snap woke up to emails telling them they no longer had jobs. By the afternoon, Snap’s stock had risen nearly 8%.
That sequence — mass layoffs, market celebration — is not new in tech. But something about Snap’s announcement felt different from the rounds of cuts that came before it. Different from the 2022 correction. Different from the 2023 efficiency drives. Different even from last year’s “rightsizing.” Because this time, Snap’s CEO Evan Spiegel didn’t just say the company was cutting costs. He said exactly what was replacing the people it was letting go.
“Rapid advancements in artificial intelligence enable our teams to reduce repetitive work, increase velocity, and better support our community, partners, and advertisers.”
One in six Snap employees. Gone. Not because the business was failing — Snap projected revenue of $1.5 billion for Q1, up 12% annually, and Spiegel said the cuts would save more than $500 million in annualized costs. Five hundred thousand dollars per person, replaced by AI tools that don’t draw a salary. The market’s reaction to that math was enthusiastic and immediate.
But zoom out, read the actual data across every source — Q1 earnings, the Stanford AI Index 2026, LinkedIn’s Labor Market Report, Layoffs.fyi, GitHub productivity studies — and the picture is genuinely more complicated than either the tech-optimist or tech-catastrophist narratives suggest. Let’s go through the numbers one by one.
📊 The AI & Developer Job Market — Key Numbers (2026)
The Snap Moment — The First Company to Say It Out Loud
What made the Snap layoff different wasn’t the number — 1,000 employees represents about 16% of its global workforce, a significant but not unprecedented cut. What made it different was the candor. Snap is almost certainly not the first company to reduce headcount because AI is handling work that previously required humans. But it may be the first company of its size to say so explicitly, in an investor letter, on the record, without euphemism.
The math that Spiegel presented to investors is the math every CEO with a finance team is now running: $500 million in annual cost savings from 1,000 people. That’s $500,000 per eliminated role — salary, benefits, equity, office space, management overhead, HR overhead — all replaced by AI tool subscriptions that cost between $20 and $200 per seat per month. The ROI calculation, at those numbers, takes approximately three to six months to become favorable.
Snap also closed more than 300 open roles — positions that had been budgeted and were being actively recruited. Those jobs aren’t on pause. They’re cancelled. The headcount that was supposed to grow those teams is not coming back. This is the detail that matters more than the layoff count. Companies aren’t just cutting current staff. They’re stopping hiring in categories where AI can do the work.
Snap is not alone. The same week, Oracle was cutting 20,000–30,000 employees and explicitly redirecting that capital to AI infrastructure. Amazon had announced 16,000 cuts across two rounds. Meta was preparing 8,000 layoffs alongside its $145 billion AI capex raise. Microsoft restructured, simultaneously cutting while reporting that its AI business hit $37 billion in revenue — up 123% year-over-year. What Snap did was name the pattern. Every company was following the same logic.
Q1 2026 Layoffs: The Scale in Full
Snap was the most-covered story, but it wasn’t the biggest. According to Layoffs.fyi, over 92,000 tech workers were laid off in Q1 2026 alone — adding to a total approaching 900,000 since 2020. Challenger Gray & Associates estimates that roughly 47.9% of 2026’s tech cuts are directly attributable to AI-driven restructuring rather than the financial conditions (post-COVID overhiring correction, interest rate pressure) that drove earlier waves.
The major Q1 headliners: Oracle (20,000–30,000), Amazon (16,000 across two rounds), Meta (8,000 announced for May), Microsoft (layoffs across several units alongside 40% Azure growth), Atlassian (1,600 — but simultaneously hiring 800 specifically for AI/MLOps). Snap’s 1,000 was small in absolute terms. In percentage terms — 16% of total headcount — it was among the most aggressive cuts in the industry.
The roles being cut first follow a clear pattern, consistent across all these companies: customer support agents, QA engineers, mid-level software engineers, data entry, technical writers, recruiters, and product managers at the execution layer. These are roles characterized by three things: high volume, defined workflows, and the kind of multi-step task integration that AI agents — like the new Codex, Claude Opus 4.7, or Cursor — can now handle in production. A QA engineer running regression tests. A technical writer translating PRD requirements into documentation. A data entry specialist moving structured information between systems. These are not judgment-intensive roles. They are execution roles. And in 2026, AI executes.
| Company | Q1 2026 Cuts | AI Attribution? | AI Investment Simultaneously |
|---|---|---|---|
| Oracle | 20,000–30,000 | Yes — explicit | Redirecting $8–10B to AI infrastructure |
| Amazon | 16,000 (two rounds) | Yes — AI efficiency cited | $105B AI capex commitment 2026 |
| Meta | 8,000 (May start) | Yes — “AI pivot” framing | $125–145B capex, MTIA custom silicon |
| Snap | ~1,000 (16% workforce) | Yes — named explicitly | $500M annual savings redirected to AI tools |
| Atlassian | 1,600 | Partial — restructuring | Hired 800 specifically for AI/MLOps |
| Salesforce | 4,000 (Sep 2025) | Yes — Benioff: “I need less heads” | Agentforce platform launch |
The Code Generation Numbers — What Companies Are Actually Reporting
The most striking data point in this entire conversation comes not from a layoff announcement but from an earnings call. During Q1 2026 results, Sundar Pichai told investors that 75% of all new code at Google is now AI-generated and approved by human engineers. Up from 25% in 2025. Up from 50% last fall. In three quarters, the majority of Google’s engineering output shifted from human-first to AI-first with human review.
Snap’s CEO reported that 65% of new code at Snap is AI-generated. Meta has targeted 55%+ agent-assisted code changes. These are not pilot programs. These are production numbers from companies shipping consumer products to billions of users. The shift from “AI helps developers write code faster” to “AI writes most of the code and developers review it” happened in the span of about eighteen months.
The implications compound. If 75% of code is AI-generated, the work that previously justified a team of 20 engineers might now require a team of 8 — with the remaining 12 positions eliminated not through layoffs but through attrition and hiring freezes. Snap’s closed job postings are the clearest evidence of this: 300 roles that were budgeted, approved, and being recruited simply cancelled. Not deferred. Cancelled. The model — AI coding agent plus human reviewer — requires fewer humans than the model it’s replacing.
The Claude Opus 4.7 review we published documented exactly this dynamic: Cursor’s internal benchmark shows Opus 4.7 clearing 70% of tasks versus 58% for Opus 4.6. Rakuten saw it resolve 3x more production tasks. These are not benchmark numbers. These are real engineering tasks, resolved autonomously, that would previously have occupied a human engineer’s full attention.
The Entry-Level Crisis — Stanford’s Most Important Finding
The Stanford HAI AI Index 2026, published April 13, contains a data point that should be required reading for every computer science student and every hiring manager in the industry: employment for software developers aged 22–25 has fallen nearly 20% since 2024. Older developers’ headcount continues to grow. The impact is entirely concentrated at the entry level.
Stanford Professor Jan Liphardt told the Los Angeles Times that Stanford CS graduates are “struggling to find entry-level jobs” — a “dramatic reversal from three years ago.” This year’s graduating class is facing one of the most difficult recruiting seasons in modern history, and the data from every direction points to the same structural cause: the work that entry-level engineers were hired to do — writing first drafts of code, handling straightforward tickets, running tests, building simple features — is now being done by AI tools that senior engineers and product managers can deploy directly.
The systemic implication is deeper than the immediate job market pain. AI is eating the bottom rung of the career ladder. Fresh graduates struggle to get hired. But those 3–5 years in are relatively safe, because they’re needed to review and oversee AI output. The long-term problem the Stanford report flags is structural: if entry-level jobs disappear, where do tomorrow’s mid-senior hires come from? The career development pipeline for software engineers has always been apprenticeship — juniors learn by doing, under the supervision of seniors. Remove the junior tier and the pipeline that produces the next generation of senior engineers doesn’t exist.
The Stanford AI Index 2026 is explicit about the timing: “Productivity gains from AI are appearing in many of the same fields where entry-level employment is starting to decline.” These two facts — productivity up, junior employment down — are not coincidental. They are the same economic event, described from two different perspectives.
What the Productivity Data Actually Shows
The productivity numbers that AI companies cite are real. GitHub Copilot studies consistently show developers completing tasks approximately 55% faster with AI assistance — roughly 1 hour 11 minutes versus 2 hours 41 minutes for comparable tasks. The Stanford AI Index 2026 compiles multiple independent studies and finds 14–26% measured productivity improvements in software development workflows. That’s not a projection. That’s observed output data from real engineering teams.
But the productivity gains are uneven, and the unevenness is precisely where the story gets complicated. The clearest gains appear in structured, high-volume, measurable work: code completion, regression testing, documentation generation, boilerplate creation. The gains are smaller — and sometimes negative — in tasks requiring deeper reasoning, architectural judgment, or domain expertise in ambiguous problem spaces.
One striking counter-data point from the Stanford report: open-source developers using AI assistance actually became 19% slower. The interpretation isn’t that AI is bad for those developers. It’s that the tasks open-source developers work on — ambiguous, architecturally complex, deeply context-dependent — are precisely the category where AI coding tools add overhead rather than throughput. The model generated code that needed to be carefully reviewed, didn’t match the project’s conventions, or solved the wrong problem efficiently. Reviewing and correcting that output took longer than writing it from scratch.
This points to the real productivity story: AI coding tools are transforming the work that the most expendable roles did. They are not (yet) transforming the work that the most valuable engineers do. The gap between “AI handles the execution” and “AI handles the architecture” remains wide — and it’s exactly the gap that separates a junior engineer’s day from a senior engineer’s day. Which is precisely why the data shows junior employment falling while senior headcount grows.
1.3 Million New Jobs — The Creation Side of the Equation
The layoff headlines dominate, but they’re only half the story. LinkedIn’s 2026 labor market data, reported to the World Economic Forum in January 2026, documents that AI has added 1.3 million new roles in just two years — AI Engineers, Forward-Deployed Engineers, Data Annotators, MLOps Engineers, AI Safety Researchers, and AI Product Managers. More than 600,000 new data center jobs have been created to support the infrastructure buildout that underpins AI’s growth.
The Stanford AI Index 2026 adds the most jaw-dropping number in this entire category: agentic AI job postings grew 10,854% year-over-year. That’s not a typo. The demand for people who can build, deploy, govern, and optimize AI agent systems — the Codex-style and Gemini Enterprise Agent Platform implementations we’ve been covering — is exploding at a rate that makes every other job category look static. AI governance roles grew 17%. Senior engineering roles grew as well, driven by demand for the humans who review and direct AI-generated output.
The pattern the data reveals is a restructuring rather than a shrinking: AI is eliminating the execution tier and creating an oversight and architecture tier. The jobs being eliminated are characterized by volume, repetition, and defined workflow. The jobs being created are characterized by judgment, governance, strategic direction, and deep AI fluency. They are not the same jobs. They do not require the same skills. And they are not accessible to the same people — a recent graduate whose CS degree prepared them to write CRUD applications is not automatically qualified for an agentic AI orchestration role without significant upskilling.
India: The Counter-Narrative Nobody Is Covering Well
The dominant Western media narrative about AI and employment focuses almost entirely on US tech layoffs. The LinkedIn AI Labor Market Report 2026, published April 24, tells a dramatically different story when you zoom out to the global picture — and India is the most important data point in it.
AI engineering hiring in India grew 59.5% year-on-year — the highest growth rate of any market in the study, ahead of the US, UK, France, and Germany. Bengaluru now matches San Francisco with 3.0% of its LinkedIn members identified as AI engineers — a parity that would have seemed implausible five years ago. Hyderabad posted 51% growth. Vijayawada, a city most Western tech observers would struggle to find on a map, posted 45.5% growth — a data point so striking that it led multiple analysts to double-check the figures.
In manufacturing specifically, AI engineering talent in India has expanded four times over, reaching 2.0% of the sector’s workforce in 2025. The growth is no longer confined to software services firms and Bangalore tech parks. It’s reaching industrial operations, small businesses, and enterprise workflows across the country.
The mechanism driving this is straightforward but important: AI tools dramatically reduced the cost of building AI-powered applications. A startup in Hyderabad with three engineers and access to GLM-5.1’s $1/million token API or Gemma 4 under Apache 2.0 can now build products that would have required a team of 20 engineers and significant cloud infrastructure investment just two years ago. The open-source model wave we documented in April directly democratized AI capability — and markets with large pools of engineering talent and relatively lower labor costs are capturing that democratization faster than markets where engineering salaries are higher.
The 59.5% India figure is not just a positive data point about India. It’s evidence that the global AI job market is not zero-sum. Job creation in AI engineering is happening at scale — it’s happening in different geographies, at different skill levels, and in response to different cost and capability structures than the market that’s being disrupted.
Is AI a Convenient Scapegoat? The Dissenting View
The Stanford Review — Stanford University’s student newspaper — published a piece this month arguing that AI is “a convenient scapegoat” for entry-level job market contraction that is actually being driven by financial conditions. The piece notes that a 2025 NBER paper studying 25,000 workers across 7,000 workplaces found precisely zero effect on earnings or hours in any occupation from AI adoption — and that the study even replicated the decline in early-career employment, showing it was not driven by firms actually adopting AI at the time of the data collection.
The counterargument: Big Tech hired millions of employees when interest rates were near zero (2020–2022). Meta grew from 45,000 to 86,000 employees in three years. Alphabet from 119,000 to 190,000. The fastest tightening cycle in 40 years followed, and every hire became subject to a much higher bar. Professor Eric Roberts at Stanford documented, in a different historical context, that similar entry-level job market contractions in computer science had zero correlation with technology displacement — they correlated with economic cycles.
This is a legitimate argument that deserves honest engagement. The 2025 NBER paper is real. The zero-rate overhiring and subsequent correction is real. The coincident timing of AI adoption and entry-level job contraction does not by itself prove causation.
However, the 2026 data that wasn’t available when that NBER paper was written is harder to explain away. Google reporting 75% of new code as AI-generated is not a lagging economic indicator. Snap explicitly citing AI as the reason for 1,000 layoffs is not post-hoc rationalization. The 10,854% growth in agentic AI job postings is not a coincidence with interest rate changes. The evidence that AI is now a material driver of employment restructuring — not the only driver, not an overnight replacement, but a genuine structural force — has strengthened considerably in 2026 relative to 2025. The honest reading is that both things are true: macro financial conditions contributed to the entry-level hiring freeze, and AI is now independently sustaining it.
What Developers Should Actually Do — The Practical Answer
Strip away the ideology on both sides and the data points to a clear set of actions for anyone building a software engineering career in 2026.
The roles that are shrinking are defined by execution at the junior tier. Writing CRUD endpoints. Maintaining CI pipelines. Running manual QA. Creating boilerplate documentation. If the primary value you add is speed of execution on defined tasks, AI can do that work at a fraction of your cost. This is not an insult. It is an honest description of what the data shows. Plan accordingly.
The roles that are growing require oversight, architecture, and AI fluency. Agentic AI job postings grew 10,854% year-on-year. AI governance roles grew 17%. Senior engineering roles that review and direct AI output are stable and growing. The common thread: humans who understand what AI can and cannot do, can specify tasks clearly enough for AI systems to execute them, and can evaluate whether the output is correct. The Stanford AI Index finding that open-source developers became 19% slower with AI tools suggests that using AI effectively is itself a skill that takes time to develop — and those who develop it well are more valuable, not less.
The geography of opportunity has changed. India’s 59.5% hiring growth is a direct invitation. For engineers in markets where AI adoption creates cost pressure on existing roles, markets with fast-growing AI infrastructure needs represent genuine opportunity — particularly given that open-source models and AI design tools have dramatically reduced the cost barrier to building AI-powered products in those markets.
Upskilling in applied AI is now mandatory, not optional. LinkedIn reports a 92% year-over-year increase in the share of learning time spent on AI-related courses. The skills in highest demand: AI Agents, Azure AI Studio, Intelligent Agents, Automated Feature Engineering, AI Productivity. These are applied, deployment-focused skills — not research-focused. The engineers who learn to orchestrate and direct AI agents, rather than competing with them, are the ones whose employment data looks like senior developers rather than junior ones.
- Claude Opus 4.7 Review: The AI That Finally Does the Hard Stuff — the model clearing 70% of real engineering tasks
- OpenAI Codex: The Developer Super App — parallel background agents, agentic engineering workflows
- April 2026: The Biggest Month Ever for Open-Source AI — the models democratizing AI-powered development globally
- Big Tech Q1 2026 Earnings — the $665B infrastructure spending that funds both the layoffs and the new roles
- GLM-5.1 Review — $1/million token AI at #1 SWE-Bench Pro, enabling low-cost AI development anywhere
- Claude Design Review — what happens when design work faces the same AI disruption as coding
- Project Glasswing — how AI is creating entirely new categories of specialized security engineering work
🧮 How Much Does AI Actually Cost Per Task?
Use our free AI Pricing Calculator to understand the real token economics behind AI coding tools — and why the “replace 1,000 engineers” math is as compelling as it is to every CFO in tech.
Try the Free AI Pricing Calculator →Compare Claude, GPT, Gemini, Qwen, and open-source model costs in real time
❓ Frequently Asked Questions
Is AI actually replacing software developers in 2026?
Partially and unevenly. Entry-level software developer employment for workers aged 22–25 has fallen ~20% since 2024, per the Stanford AI Index 2026. 75% of Google’s new code and 65% of Snap’s new code is now AI-generated. But senior developer headcount continues to grow, and 1.3 million new AI-related roles have been created globally in the last two years. AI is restructuring the developer job market rather than shrinking it overall — eliminating execution roles at the junior tier while creating oversight, architecture, and AI engineering roles above it.
Why did Snap’s stock rise after laying off 1,000 people?
Snap projected $500M+ in annualized cost savings from the 1,000 layoffs while maintaining or improving product output through AI tools. Investors rewarded the improved unit economics: the same or better output at significantly lower cost. CEO Evan Spiegel explicitly cited AI-driven efficiencies as enabling smaller teams to ship faster. The stock rose ~8% on the announcement.
How many tech workers were laid off in Q1 2026?
Over 92,000 tech workers were laid off in Q1 2026, according to Layoffs.fyi — approaching a total of 900,000 since 2020. Challenger Gray & Associates estimates approximately 47.9% of 2026’s tech cuts are directly attributable to AI-driven restructuring, versus prior years where financial conditions (overhiring correction, interest rate pressure) were the primary driver.
What jobs are AI creating in 2026?
LinkedIn and WEF data shows 1.3 million new AI-related roles created in two years: AI Engineers, Forward-Deployed Engineers, Data Annotators, MLOps Engineers, AI Governance roles, AI Product Managers, and more than 600,000 new data center jobs. Agentic AI job postings grew 10,854% year-over-year per the Stanford AI Index 2026. Head of AI positions are growing rapidly across Australia, Canada, India, Germany, the UK, and the US.
Why is India’s AI hiring growing so fast?
India’s AI engineering hiring grew 59.5% year-on-year — the highest of any market in LinkedIn’s 2026 report. Drivers include a large English-speaking engineering talent base, lower labor costs that make AI-powered product development highly economically attractive, rapid adoption of AI tools in manufacturing and SMBs, and the democratizing effect of open-source models that reduce the cost barrier to building AI products. Bengaluru now matches San Francisco in AI engineer density at 3.0% of LinkedIn members.
What should developers do to stay relevant in an AI-driven market?
Focus on skills that complement rather than compete with AI: agentic AI orchestration, AI governance, architectural decision-making, AI output review and quality assurance, and applied AI tool deployment. The Stanford AI Index documents that engineers who use AI tools effectively are significantly more productive — the skill is learning to direct AI systems well, not competing with them on execution speed. Skills with the highest demand growth include AI Agents, Azure AI Studio, Intelligent Agents, and Automated Feature Engineering.
Latest Articles
Browse our comprehensive AI tool reviews and productivity guides
Musk v. OpenAI Trial: The Case That Could Reshape the Entire AI...
Musk v. OpenAI Trial: The Case That Could Reshape the Entire AI Industry
Musk called himself "a fool" on the stand. Altman appeared by prerecorded video from AWS while being sued. The judge reprimanded both sides. And the AI industry's most consequential legal battle is just getting started.
Big Tech Q1 2026 Earnings: The $665 Billion AI Bet — Winners,...
Big Tech Q1 2026 Earnings: The $665 Billion AI Bet — Winners, Losers, and What It Means
Five tech giants reported Q1 2026 earnings in 48 hours. Combined AI capex: $665 billion — 75% more than 2025. Alphabet and Amazon won. Meta spooked investors. Here's every number that matters.
AI Is Replacing Developers — The Real Numbers (2026)
AI Is Replacing Developers — The Real Numbers (2026)
Snap fired 1,000. Google generates 75% of new code with AI. Entry-level developer jobs fell 20%. But 1.3M new AI roles were created and India's AI hiring surged 59.5%. Here's what's actually happening.
I Used Claude Free for 3 Months Instead of ChatGPT and Gemini...
I Used Claude Free for 3 Months Instead of ChatGPT and Gemini — Here’s What Happened
I launched and grew NivaaLabs on Claude's free tier for 3 months. I also used ChatGPT and Gemini. Here's the honest, task-by-task breakdown of what each AI actually does well — and which one I'd recommend for someone building something real on a $0 budget.
Runway Gen-3 Turbo: Real-Time Video Tested (2026)
Runway Gen-3 Turbo: Real-Time Video Tested (2026)
Runway Gen-3 Turbo's real-time video generation capabilities are put to the test, examining quality, speed, and value.
Best AI Coding Tools 2026: Every Major Tool Ranked — Cursor, Claude...
Best AI Coding Tools 2026: Every Major Tool Ranked — Cursor, Claude Code, Copilot, Windsurf & More
85% of developers now use AI coding tools daily. AI writes 46% of all new code. The market has 10+ serious tools and most developers end up using two or three. Here's how every major AI coding tool in 2026 ranks — with real benchmark data, honest pricing, and a verdict for every workflow type.
DeepSeek V4 Review: V4 Flash & V4 Pro — Almost Frontier, a...
DeepSeek V4 Review: V4 Flash & V4 Pro — Almost Frontier, a Fraction of the Price (April 2026)
DeepSeek V4 arrived April 24, 2026 — one year after R1 shook Silicon Valley. V4 Pro is the world's largest open-weight model at 1.6 trillion parameters. V4 Flash is cheaper than GPT-5.4 Nano. And both run on Chinese chips. Here's everything you need to know.
GPT-5.5 vs Claude Opus 4.6 (2026): Which AI Model Wins for Your...
GPT-5.5 vs Claude Opus 4.6 (2026): Which AI Model Wins for Your Work?
OpenAI's GPT-5.5 arrived April 23 claiming to be the smartest model yet. Anthropic's Claude Opus 4.6 still holds the top Chatbot Arena ELO. Both cost real money. Which one actually wins for your workflow? Here's the full data-driven comparison.
GPT-5.5 Review: OpenAI’s Smartest Model Yet — Agentic Coding, Computer Use &...
GPT-5.5 Review: OpenAI’s Smartest Model Yet — Agentic Coding, Computer Use & More (April 2026)
GPT-5.5 landed April 23 — seven weeks after 5.4. OpenAI calls it a "new class of intelligence for real work." It's faster per token, stronger at agentic coding, computer use, and scientific research, and comes with the strongest safety guardrails yet. Here's everything you need to know.
Project Glasswing: Anthropic’s “Too Dangerous to Release” AI and the Cybersecurity Reckoning
Project Glasswing: Anthropic’s “Too Dangerous to Release” AI and the Cybersecurity Reckoning
Anthropic built an AI so capable at hacking that they won't release it publicly. Claude Mythos Preview found a 27-year-old OpenBSD zero-day for under $50. Project Glasswing is what happens next.
Google Cloud Next 2026: Every Major Announcement — Agents, TPU 8, Virgo...
Google Cloud Next 2026: Every Major Announcement — Agents, TPU 8, Virgo Network & More
Google Cloud Next 2026 just happened. Here's everything: new 8th-gen TPUs, the Gemini Enterprise Agent Platform, A2A protocol in production at 150 orgs, Workspace Studio for no-code agents, and a $185B infrastructure bet. One article, all the details.
OpenAI ChatGPT Ads Review: The $100 Billion Bet That’s Already Getting Messy
OpenAI ChatGPT Ads Review: The $100 Billion Bet That’s Already Getting Messy
OpenAI launched ads in ChatGPT, faced user backlash, fired back at Anthropic's Super Bowl jabs, and just flipped to cost-per-click pricing. Here's what's actually happening.