Edition, Thu, 05 Nov 2026.

Big Tech's AI buildout is still accelerating

Record capex from Amazon, Microsoft, Alphabet, and Meta left Wall Street waiting for the returns to show up.

Record capex from Amazon, Microsoft, Alphabet, and Meta left Wall Street waiting for the returns to show up.

After the Q3 earnings calls from the four biggest AI spenders, the message was consistent: the data center buildout is not slowing. The four companies plan at least $720 billion in capital spending this year, with more than $1 trillion in additional commitments already arranged for coming years. Alphabet's CFO Anat Ashkenazi said spending would rise "significantly" in 2027, and Amazon CEO Andy Jassy called the buildout a "once in a lifetime opportunity" to grow the company to a $1 trillion business.

The payoff, though, remains elusive. Google's search ad growth slowed from 17% a year ago to 12% as AI Mode queries monetize less, and Alphabet's stock posted its worst one-day drop in history, falling 13%. Microsoft, previously the only hyperscaler avoiding debt for the buildout, announced a $25 billion bond sale shortly after its call. Anthropic's own numbers show the cost structure: $17 billion in revenue against $17.5 billion in compute costs, plus nearly $900 billion in future infrastructure commitments.

Data centers are losing the political battle

After the midterms, new Democratic majorities hold the levers to put the AI buildout on pause.

After the midterms, new Democratic majorities hold the levers to put the AI buildout on pause.

The November 3 midterms produced a clear mandate against data center expansion. Exit polls found almost 80% of voters opposed to a data center in their neighborhood, nearly 60% cited data centers as a crucial or major factor in their vote, and more than half said AI will do more harm than good. In Ohio, the county where Meta planned a 700,000-square-foot facility passed a ban on data centers, and Democrats' full control of Congress opens the door to a nationwide moratorium proposed by Bernie Sanders and AOC.

Texas, the fastest-growing data center market in the US, saw the sharpest blue wave: James Talarico defeated Ken Paxton for the open Senate seat, Gina Hinojosa beat Governor Greg Abbott, and Democrats took the state house for the first time in 24 years. Hinojosa is pushing a statewide moratorium, putting mega-projects like the 1,820-megawatt Stargate Frontier in Shackelford County and Anthropic's 6,400-megawatt Nexus Hubbard campus, the second-largest data center being built in the country, in question.

An Anthropic flaw that won't get fixed

A browser extension can run code on your machine through Claude Dispatch, and Anthropic says it's not fixing it.

A browser extension can run code on your machine through Claude Dispatch, and Anthropic says it's not fixing it.

Gal Weizman of Forever Security found a vulnerability he dubbed "ThisBash" in Claude Dispatch: any browser extension that can reach claude.ai can send commands to Claude Desktop on a user's computer, or on another machine signed into the same account, with as little as a one-second auto-approve click. In his proof of concept, a script in the page's JavaScript context installed a malicious npm package through a preinstall hook, achieving remote code execution on the host machine.

Anthropic acknowledged the report in May, closed it as a duplicate of a known flaw, paid no bug bounty, and issued no CVE. Dispatch was slated to be retired by the end of October but remains available in limited beta, and Weizman says the design pattern persists in other features like Claude Code Remote Control and Claude Cowork. With roughly 950 million monthly visits to claude.ai and personal Claude accounts ranging from 18% to 71% of seats across six surveyed organizations, a compromised browser on a personal laptop can act as a remote entry point into a corporate network.

OpenAI acquires Pierre Computer to build out Codex

The deal folds the Code Storage, Trees, and Diffs libraries into OpenAI's coding agent platform.

The deal folds the Code Storage, Trees, and Diffs libraries into OpenAI's coding agent platform.

OpenAI said Wednesday it will acquire Pierre Computer Company, the maker of the Code Storage, Trees, Diffs, and Highlights open-source libraries, to enhance its agentic coding platform Codex with distributed storage capabilities. Financial terms were not disclosed, and the deal awaits customary regulatory approvals. Trees and Diffs already power file browsing and change review in Codex, and Code Storage will join them to expand the platform's tools for building, understanding, and reviewing code.

Discord is going all-in on Rust

Ditching Electron, Discord's new desktop app idles at 290MB of RAM where the old build could burn a gigabyte.

Ditching Electron, Discord's new desktop app idles at 290MB of RAM where the old build could burn a gigabyte.

Discord chief technology officer Stanislav Vishnevskiy said the company has been an early Rust adopter since 2018, and moving the full client to Rust means Electron goes away, as does React Native. Mobile keeps its native UI but will share app code with desktop via Rust, and Vishnevskiy shared an image of the Windows client using just 290MB of RAM, a sharp improvement over the current app, which can easily hit 1GB or more once you join a few servers. A full switch to Rust is expected to enter testing with users soon.

ServiceNow and LTM push agentic AI past the pilot phase

The two companies have expanded their partnership to scale governed agentic AI deployment for more than 700 shared customers.

The two companies have expanded their partnership to scale governed agentic AI deployment for more than 700 shared customers.

ServiceNow and LTM announced on November 5 a planned expansion of their partnership to scale agentic AI and accelerate joint go-to-market initiatives. The companies will combine the ServiceNow AI Platform with LTM's domain, engineering, and transformation expertise to help enterprises move agentic AI programs from isolated use cases toward governed, enterprise-wide adoption. LTM, an Elite ServiceNow Partner, will serve as a "client-zero" reference case, using its own internal transformation as the blueprint for joint solutions and go-to-market programs.

LTM has deployed 1,500 ServiceNow AI Agents across its operations, creating a digital workforce that works alongside its employees, with the ServiceNow Super Intelligence Control Tower providing centralised oversight of AI inventory and risk. LTM expects up to an 80 percent improvement in AI asset management efficiency, and its existing deployment is already delivering measurable outcomes across cybersecurity, HR, IT, and project delivery, including up to 87 percent faster senior management review cycles and reduced IT ticket volumes. Kulmeet Singh Bawa, GVP and MD of ServiceNow India, framed governance as the central issue, noting that each AI agent is a new enterprise identity requiring visibility and accountability.

ServiceNow stock tests support after a bruising year

Shares down 40.6% over the past 12 months are sitting at a price level they have bounced off before, with key earnings details in view.

Shares down 40.6% over the past 12 months are sitting at a price level they have bounced off before, with key earnings details in view.

ServiceNow shares have lost 40.6% over the past year, against an 11.9% gain for the S&P 500, with the sharpest single-day decline on October 29, 2026, when the stock fell 27 percent. At $109.16, shares have returned to a level reached several times since 2023, which Trefis notes has held three times before. A $10,000 investment at today's price would grow to about $12,450 at the $135.86 resistance level, while support is broken below at $103.70.

The business is growing slightly below its historical pace: revenue rose 21.8% over the twelve months to September 30, 2026, just under the three-year average of 22.1%, and operating margin fell to 9.7% from 13.9% a year earlier. On its October 28 earnings call, management lifted its goal for annual contract value from AI to $1.7 billion by the end of 2026, up from the $1.5 billion target set in May, said EmployeeWorks reached $100 million of annual contract value in under three quarters, and reported Asset Management crossed $1 billion in ACV. The stock now trades at 7.3 times sales, close to the lowest quarter-end multiple since 2016.

Open-source ITSM add-ons are everywhere and nowhere

A new analysis argues the Znuny, OTOBO, and Zammad plugin catalog is not missing, it is scattered.

A new analysis argues the Znuny, OTOBO, and Zammad plugin catalog is not missing, it is scattered.

Tobias Bück, founder of Open Ticket AI and maintainer of Open ITSM Hub, argues that open-source ITSM add-ons are hard to find not because they don't exist, but because there is no shared, trustworthy catalog. Packages hide in at least six places: vendor and project shops, GitHub and GitLab repositories, the legacy OTRS Package Archive, forums and mailing lists, private internal forks at consultancies, and comparison sites. The result is that discovery is fragmented and easy to misread, which he says wastes days of operator time and "quietly kills otherwise good migrations."

The OTRS-family forks (Znuny, OTOBO, KIX) share the .opm package format, but compatibility is more fragile than it looks: framework version filters reject packages pinned to older versions, runtime failures come from renamed Perl modules, changed cache layers, and new frontend architectures, and KIX 18 is the most divergent fork. Bück's prescription is not a new closed marketplace but "shared, boring metadata" such as last-tested versions, license files, and abandonment statements, treating discovery as part of service design.