Top 10 AI News — October 30, 2026

Big Tech reported this week, and the market made clear that AI revenue growth is no longer enough unless the cash comes back too. At the same time, developer tooling kept moving: GitHub and Google both rebuilt their platforms around coding agents that come from many vendors.

1. Alphabet has its worst day on record as AI spending pushes cash flow negative

Alphabet shares fell nearly 13% on Thursday, CNBC reported, the stock's worst day ever, erasing about $500 billion in market value. The trigger was the third-quarter report Alphabet released on October 28.

Revenue rose 25% to $127.58 billion, and Google Cloud revenue grew 103% to $30.71 billion, according to 24/7 Wall St. But earnings per share of $2.12 missed estimates, mostly because of an $8.7 billion loss on equity investments. Capital spending reached $51.3 billion, up from $44.9 billion the quarter before, and free cash flow was negative $9.1 billion. It is the second deficit in a row, and analysts had expected about $2.7 billion.

To pay for the buildout, Alphabet issued $28.7 billion in senior unsecured debt and raised $12.4 billion through at-the-market stock sales. It kept its 2026 spending plan and said spending will rise significantly in 2027. Sundar Pichai said the company is "now on a half-trillion dollar revenue run rate." Investors cared more about the cash than the growth.

2. AWS grows 39%, its fastest in nearly five years, but Amazon's capex nearly doubles

Amazon reported third-quarter results on October 29. AWS revenue rose 39% to $46.0 billion, the sixth straight quarter of acceleration and the fastest growth in nearly five years. AWS operating income was $17.8 billion, up from $11.4 billion a year earlier, and the contracted backlog more than doubled to $578 billion.

The cost is the problem. Capital expenditures were $77.3 billion in the quarter, compared with $35.1 billion a year earlier and a FactSet estimate of $58.7 billion, GeekWire reported. Free cash flow over the past twelve months was an outflow of $42.4 billion. Net sales rose 12% to $201.9 billion, but earnings missed, and the stock fell more than 7% after hours.

Andy Jassy said AWS revenue growth should top 40% in both 2027 and 2028, and that Amazon will spend "meaningfully more" in 2027 than the roughly $220 billion planned for this year.

3. Microsoft beats on Azure but comes in under its own capex target

Microsoft reported revenue of $92.3 billion for its fiscal first quarter, up 19%, with Azure growing 48% against a 44% consensus. Commercial remaining performance obligation, the measure of signed future revenue, rose 73% to $677 billion. The stock rose about 2%.

Capital spending was $45.2 billion, below the more than $50 billion Microsoft had guided to. CNBC reported that the company blamed delayed deliveries of third-party capacity, and CFO Amy Hood said some of that capacity is now pushed into the first half of 2027.

Taken with the Alphabet and Amazon results, the gap is clear. Microsoft spent less than planned and kept $18.6 billion of positive free cash flow, and investors rewarded it. Alphabet and Amazon spent more and went cash-negative, and they were punished. Our October 28 brief covered the nearly $80 billion Meta and SpaceX raised for compute the same week.

4. GitHub says agents now start more than 40% of pull requests

At GitHub Universe on October 28, GitHub released its Octoverse report. In September, more than 40% of pull requests began as an agent-created draft, up from 17% in March, which is 110 million pull requests in a single month. GitHub calls that number a floor, because it cannot detect every agent. Developers pushed code 17.9 billion times in the twelve months to September 30, more than 2024 and 2025 combined, and the platform now counts over 225 million developers.

The product news was Copilot Fusion, which routes each step of a coding task to one model or a mix of models. Jay Parikh, Microsoft's EVP of CoreAI, wrote that Fusion Max scored within two points of Claude Opus 5.5 Max on the Artificial Analysis Coding Agent Index at half the cost, The Next Web reported. GitHub also made the Copilot app generally available. It runs agent sessions across several repositories in parallel and lets a session move between the app, VS Code, the command line and the web.

Pricing changed too. GitHub Enterprise customers can now use every Copilot feature without a separate subscription and pay only for the AI credits they consume. The load shows up in the infrastructure: GitHub added 4 million CPU cores and 160 petabytes of storage this year and is moving to a multi-region Azure deployment.

5. Android Studio opens to any coding agent

Google made "Bring Your Own Agent" stable in Android Studio Rabbit 2 on the first day of the Android Dev Summit, Tech Times reported on October 30. The IDE now accepts more than 40 coding agents that speak the Agent Client Protocol, including Anthropic's Claude Agent, OpenAI Codex and Google's Antigravity.

ACP is a JSON-RPC standard that Zed Industries published in August 2025. It is now governed by a group that includes Zed, JetBrains, GitHub, Google and Cognition. Google also shipped official Android CLI plugins for Claude Code and Codex, and said its internal tests found agents using android preview finished UI tasks 77% faster with up to three times fewer tokens.

This is the second platform owner this week to stop tying its developer tools to its own model. The editor is becoming neutral ground, and the agents compete inside it.

6. Google gives Gemini 4 Carbon to cyber defenders first

Google announced Gemini 4 Carbon on October 28, two weeks after Gemini 4 Argon, 9to5Google reported. Google says it ranks first on version 5.0 of the Artificial Analysis Intelligence Index with a score of 59, and climbed 89 Elo points to fourth on WebDev Arena.

The release order matters more than the scores. Carbon goes to a group of cyber defenders this week, before developers, businesses or consumers. Google reports 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation and 71.0% on CWE-bench for patching. The first access program, Fairwind, is limited to security teams, requires phishing-resistant login and bans reselling.

Introductory pricing is $2 per million input tokens and $10 per million output, rising later to $4 and $20. Google also says Carbon makes about half as many unsupported claims as Argon in internal testing. The October 1 brief covered Argon's launch.

7. The DOJ asks questions about Nvidia's $12.9 billion Hugging Face deal

The US Department of Justice has asked how Nvidia's $12.93 billion purchase of Hugging Face would affect competition in AI, Reuters reported on October 28, citing two people familiar with the matter. Officials want to know whether the deal would give Nvidia control of a key distribution point for AI models, and whether Nvidia could use the platform against rivals.

Reuters did not report a formal merger review, and neither Nvidia nor the DOJ confirmed the inquiries. The questions sit inside a longer DOJ investigation into Nvidia's conduct that also covers its SchedMD purchase and its Groq licensing deal.

Nvidia expects the deal to close in the first half of 2027 and has said Hugging Face will stay neutral across rival chips, clouds and software. If the DOJ tests that promise, it will be the first serious antitrust look at who owns the open-model supply chain.

8. Cresta opens its agent platform to developers

Cresta opened its customer-service agent platform to self-service development on October 29. Developers can now build, test and deploy agents from their own code repositories and CI/CD pipelines, with no dedicated AI team. The release includes Cresta Conductor, which supports the open Agent Skills standard, along with a Python SDK, a CLI and a certification program.

Cresta says some enterprise teams have gone from requirements to production in as little as one week. In one case, a developer at an unnamed Australian telecom provider rebuilt a production chat agent with more than 10 subagents in under a week. TELUS Digital and Atento are already building on it, and Cresta's existing customers include United Airlines, CVS Health and Marriott.

Contact-center vendors used to sell agents as finished services that their own engineers deployed. Handing the build to customers' developers is how they intend to scale.

9. "Botsitting": checking AI work takes as much effort as training a new hire

A Founder Reports survey published October 30 asked 2,052 US workers who use AI at least weekly what the oversight costs them. Among the 1,856 who have trained an employee, 60% said training, checking and fixing AI work takes at least as much effort as training a person. One in three said it takes more.

The time is real. 31.6% spend at least three hours a week on what the report calls "botsitting," and 8.4% spend more than ten. Managers carry more of it: 40.7% of managers and above spend at least three hours a week overseeing AI and fixing its errors, against 25% of individual contributors.

The finding matches an older academic result the report cites. Humlum and Vestergaard found in 2025 that 59% of new AI-related tasks were implementation and oversight. The productivity gain is real, but someone in the org chart pays for it in review time.

10. A quarter of employees have built an AI app they use at work

Pave, a Quickbase product, published survey results on October 28 from 5,177 US employees and 527 business leaders. 38% of employees had used AI to build an app or working prototype in the past year, and 24% built one they now use for work. Of those builders, 84% replaced a process that had run on spreadsheets, email, chat or paper.

Leaders mostly approve: 83% who found an employee-built app called it important or essential, and fewer than 2% ordered anyone to stop. But 53% of organizations found at least one app that IT and leadership did not know about. 57% of leaders require security reviews, while only 32% of builders have access to someone who can perform one. Pave sells a product aimed at this gap, so read the numbers with that in mind.

What to watch

The Wall Street Journal has reported that Anthropic is targeting the week of November 16 for its listing, which lands in a market that has just punished Alphabet and Amazon for cash burn. Watch whether investors judge the S-1 on revenue growth or on free cash flow. For developers, watch whether JetBrains and Apple follow GitHub and Google toward agent-neutral tooling, and how many teams take up GitHub's credit-only Copilot pricing once it goes self-serve.