AI and IT News Recap: July 7, 2026: Anthropic's 9 Billion Kentucky Data-Center Bet, Microsoft Cuts Nearly 5,000 More Jobs, and an Opera GX Flaw That Reads Your ScreenBy Noah Smith, Owner & Consultant, KeyChange Technologies · July 7, 2026
Here is the AI and IT news for July 7, 2026, a fast, no-spin read for business owners. Model launches stayed quiet, but the money and the risk did not. Anthropic signed a
9 billion, 20-year lease to build out AI compute in rural Kentucky, Microsoft cut nearly 5,000 more jobs while it pours billions into AI, and researchers showed how a single visit to a bad website could silently plant a browser add-on. The most useful items sit right at the top.
📌 The AI and IT news at a glance
- Anthropic signed a 9 billion, 20-year data-center lease with TeraWulf to power its AI in Hawesville, Kentucky.
- Microsoft laid off around 4,800 people, hitting Xbox hardest, even as it spends heavily on AI.
- An Opera GX flaw let any website silently install a browser mod and lift your signed-in Gmail address with no click.
- Tokyo police arrested a 15-year-old who used AI-assisted code to cancel roughly 46,800 streaming accounts.
🔝🤖 Top story: Anthropic's 9 billion bet on a Kentucky data center
Anthropic signed a 20-year lease with bitcoin-miner-turned-data-center-operator TeraWulf for a purpose-built AI campus at TeraWulf's Justified Data site in Hawesville, Kentucky, a deal expected to generate about
9 billion in contracted revenue over the initial term. The campus is designed to deliver roughly 401 megawatts of critical IT load, with the first capacity expected online in the second half of 2027 and full capacity by early 2028. Anthropic holds two five-year renewal options that could stretch the relationship to 30 years, and TeraWulf separately agreed to sell its majority stake in a Texas data-center joint venture to a group led by Fluidstack for about $530 million.
The market reaction told the story: TeraWulf shares jumped more than 16% before the opening bell. This is the AI infrastructure buildout in miniature, a frontier lab locking up two decades of power and space in a small river town so it can keep training and serving models. The scale is enormous, and so is the bet that demand will still be there when the lights come on in 2027.
In short: Anthropic signed a 20-year, roughly
9 billion lease with TeraWulf for a 401-megawatt AI data center in Hawesville, Kentucky, with initial capacity due in late 2027.
What it means for your business: The AI tools you rely on increasingly depend on multi-billion-dollar, multi-year infrastructure commitments, which points to serious staying power but also to providers that need massive, sustained demand to justify the spend.
My take: Deals like this are why your AI subscriptions keep getting more capable. They are also a reminder that a handful of firms are making twenty-year, capital-heavy wagers on a technology that changes every few months. For your own planning, treat frontier AI as durable infrastructure worth building on, but keep your data portable and avoid wiring a critical process so tightly to one vendor that you cannot move if pricing or priorities shift.
Source: CNBC
🤖 AI generally
🤖 Microsoft cuts nearly 5,000 jobs while AI spending climbs
Microsoft laid off around 4,800 people on Monday, about 2.1% of its global workforce, with the Xbox division hit hardest at roughly 1,600 roles and about 3,200 gaming cuts expected in total through fiscal 2027. Xbox chief Asha Sharma called it "the most significant restructure in Xbox history" and said bluntly that the business "is not healthy," operating at margins far below comparable platforms. Chief people officer Amy Coleman told staff the eliminated roles "are not being replaced by AI," while conceding that "AI is changing how work gets done."
The timing is hard to ignore. The cuts land just days after Microsoft launched its Frontier Company unit, a .5 billion push to deploy enterprise AI, and they fit a pattern across Big Tech this year where layoffs and soaring AI investment move together. Roughly 154,000 tech workers have lost jobs in the first half of 2026, with Meta, Oracle, and Amazon among the names cutting deep.
In short: Microsoft cut about 4,800 jobs, roughly 2.1% of staff, hitting Xbox hardest, days after committing .5 billion to a new enterprise AI unit.
What it means for your business: When the largest software vendors trim headcount while raising AI spend, expect them to keep pushing AI-assisted tooling into the products you buy, and expect "do more with fewer people" to become the default sales pitch.
My take: Ignore the semantic fight over whether AI "caused" these layoffs. The useful signal is that even Microsoft is reallocating dollars from people to AI infrastructure and betting it comes out ahead. For a small business, the lesson is not to slash staff, it is to be deliberate: identify the repetitive work AI genuinely handles well, free your people for the judgment calls it cannot, and measure whether the tools actually pay for themselves.
Source: TechCrunch
🛡️ IT and security
🛡️ An Opera GX flaw let websites silently install a browser mod and read your Gmail
Researchers disclosed a flaw in Opera GX, the gaming-focused version of the Opera browser, that let a malicious website silently install one of the browser's "GX Mods" add-ons and then use it to lift data from the pages a victim visited. In a proof of concept, they reconstructed a signed-in user's full Gmail address from a single visit with no click required, using advanced CSS injection rules to read data on the page and quietly beacon it back to an attacker's server. A cruder second trick in the same install path could crash the browser in private mode and dump every open tab, and that piece affects regular Opera too.
Opera rated the issue P1, its top severity, paid the maximum $5,000 bug bounty, and shipped a fix in Opera GX version 130.0.5847.89. The company says it found no evidence the flaw was ever used in the wild, so anyone on a current build is already protected. The takeaway is less about Opera specifically and more about how much power a browser add-on quietly holds over everything you do online.
In short: Opera patched a GX Mods flaw that let any website silently install a browser add-on and reconstruct a signed-in user's Gmail address with no click, with no sign it was exploited in the wild.
What it means for your business: Browser extensions and mods run with deep access to the pages your team visits, so a single bad one, or a flaw in the install process, can quietly leak account details and internal data.
My take: This is a good prompt to audit browser extensions across your team. Keep browsers set to auto-update so fixes like this land on their own, remove add-ons nobody actively uses, and be wary of niche or "fun" browser features that ask for broad permissions. The convenient extension you forgot you installed is exactly the kind of thing an attacker hopes you never think about.
Source: The Hacker News
🛡️ A 15-year-old used AI-assisted code to boot 46,800 users off a streaming service
Tokyo police arrested a 15-year-old student suspected of exploiting a vulnerability in the Bandai Channel anime streaming service to cancel roughly 46,800 other users' accounts, according to Jiji reporting carried by Nippon.com. Investigators say the teenager leaned on AI-assisted code generation to build the attack, and the case is being pieced together from server requests, traffic analysis, account-cancellation records, and IP-change history.
It is a small incident with an outsized lesson. The barrier to writing functional attack code keeps dropping as AI coding assistants improve, which means the pool of people who can cause real disruption is widening well beyond skilled professionals. The underlying failure was still an ordinary application vulnerability, the kind that proper access controls and rate limiting are meant to catch.
In short: Tokyo police arrested a 15-year-old who allegedly used AI-assisted code to exploit a Bandai Channel flaw and cancel about 46,800 user accounts.
What it means for your business: AI coding tools lower the skill needed to turn a known weakness into a working attack, so ordinary application bugs on your customer-facing systems now attract a much larger set of potential attackers.
My take: You do not need a nation-state on your threat list to get burned, you need an exposed flaw and a curious teenager with a chatbot. Make sure anything customer-facing enforces real authorization checks, rate-limits sensitive actions like cancellations and password resets, and logs enough to reconstruct what happened. The fundamentals that stop a bored 15-year-old are the same ones that slow down professionals.
Source: Nippon.com
That is the AI and IT news for July 7, 2026. For the last edition, see our July 6 recap, and browse practical how-tos in our Knowledge Base.
By Noah Smith, Owner & Consultant, KeyChange Technologies · July 7, 2026
Here is the AI and IT news for July 7, 2026, a fast, no-spin read for business owners. Model launches stayed quiet, but the money and the risk did not. Anthropic signed a
📌 The AI and IT news at a glance
- Anthropic signed a 9 billion, 20-year data-center lease with TeraWulf to power its AI in Hawesville, Kentucky.
- Microsoft laid off around 4,800 people, hitting Xbox hardest, even as it spends heavily on AI.
- An Opera GX flaw let any website silently install a browser mod and lift your signed-in Gmail address with no click.
- Tokyo police arrested a 15-year-old who used AI-assisted code to cancel roughly 46,800 streaming accounts.
🔝🤖 Top story: Anthropic's
9 billion bet on a Kentucky data centerAnthropic signed a 20-year lease with bitcoin-miner-turned-data-center-operator TeraWulf for a purpose-built AI campus at TeraWulf's Justified Data site in Hawesville, Kentucky, a deal expected to generate about
9 billion in contracted revenue over the initial term. The campus is designed to deliver roughly 401 megawatts of critical IT load, with the first capacity expected online in the second half of 2027 and full capacity by early 2028. Anthropic holds two five-year renewal options that could stretch the relationship to 30 years, and TeraWulf separately agreed to sell its majority stake in a Texas data-center joint venture to a group led by Fluidstack for about $530 million.The market reaction told the story: TeraWulf shares jumped more than 16% before the opening bell. This is the AI infrastructure buildout in miniature, a frontier lab locking up two decades of power and space in a small river town so it can keep training and serving models. The scale is enormous, and so is the bet that demand will still be there when the lights come on in 2027.
In short: Anthropic signed a 20-year, roughly
9 billion lease with TeraWulf for a 401-megawatt AI data center in Hawesville, Kentucky, with initial capacity due in late 2027.What it means for your business: The AI tools you rely on increasingly depend on multi-billion-dollar, multi-year infrastructure commitments, which points to serious staying power but also to providers that need massive, sustained demand to justify the spend.
My take: Deals like this are why your AI subscriptions keep getting more capable. They are also a reminder that a handful of firms are making twenty-year, capital-heavy wagers on a technology that changes every few months. For your own planning, treat frontier AI as durable infrastructure worth building on, but keep your data portable and avoid wiring a critical process so tightly to one vendor that you cannot move if pricing or priorities shift.
Source: CNBC
🤖 AI generally
🤖 Microsoft cuts nearly 5,000 jobs while AI spending climbs
Microsoft laid off around 4,800 people on Monday, about 2.1% of its global workforce, with the Xbox division hit hardest at roughly 1,600 roles and about 3,200 gaming cuts expected in total through fiscal 2027. Xbox chief Asha Sharma called it "the most significant restructure in Xbox history" and said bluntly that the business "is not healthy," operating at margins far below comparable platforms. Chief people officer Amy Coleman told staff the eliminated roles "are not being replaced by AI," while conceding that "AI is changing how work gets done."
The timing is hard to ignore. The cuts land just days after Microsoft launched its Frontier Company unit, a .5 billion push to deploy enterprise AI, and they fit a pattern across Big Tech this year where layoffs and soaring AI investment move together. Roughly 154,000 tech workers have lost jobs in the first half of 2026, with Meta, Oracle, and Amazon among the names cutting deep.
In short: Microsoft cut about 4,800 jobs, roughly 2.1% of staff, hitting Xbox hardest, days after committing .5 billion to a new enterprise AI unit.
What it means for your business: When the largest software vendors trim headcount while raising AI spend, expect them to keep pushing AI-assisted tooling into the products you buy, and expect "do more with fewer people" to become the default sales pitch.
My take: Ignore the semantic fight over whether AI "caused" these layoffs. The useful signal is that even Microsoft is reallocating dollars from people to AI infrastructure and betting it comes out ahead. For a small business, the lesson is not to slash staff, it is to be deliberate: identify the repetitive work AI genuinely handles well, free your people for the judgment calls it cannot, and measure whether the tools actually pay for themselves.
Source: TechCrunch
🛡️ IT and security
🛡️ An Opera GX flaw let websites silently install a browser mod and read your Gmail
Researchers disclosed a flaw in Opera GX, the gaming-focused version of the Opera browser, that let a malicious website silently install one of the browser's "GX Mods" add-ons and then use it to lift data from the pages a victim visited. In a proof of concept, they reconstructed a signed-in user's full Gmail address from a single visit with no click required, using advanced CSS injection rules to read data on the page and quietly beacon it back to an attacker's server. A cruder second trick in the same install path could crash the browser in private mode and dump every open tab, and that piece affects regular Opera too.
Opera rated the issue P1, its top severity, paid the maximum $5,000 bug bounty, and shipped a fix in Opera GX version 130.0.5847.89. The company says it found no evidence the flaw was ever used in the wild, so anyone on a current build is already protected. The takeaway is less about Opera specifically and more about how much power a browser add-on quietly holds over everything you do online.
In short: Opera patched a GX Mods flaw that let any website silently install a browser add-on and reconstruct a signed-in user's Gmail address with no click, with no sign it was exploited in the wild.
What it means for your business: Browser extensions and mods run with deep access to the pages your team visits, so a single bad one, or a flaw in the install process, can quietly leak account details and internal data.
My take: This is a good prompt to audit browser extensions across your team. Keep browsers set to auto-update so fixes like this land on their own, remove add-ons nobody actively uses, and be wary of niche or "fun" browser features that ask for broad permissions. The convenient extension you forgot you installed is exactly the kind of thing an attacker hopes you never think about.
Source: The Hacker News
🛡️ A 15-year-old used AI-assisted code to boot 46,800 users off a streaming service
Tokyo police arrested a 15-year-old student suspected of exploiting a vulnerability in the Bandai Channel anime streaming service to cancel roughly 46,800 other users' accounts, according to Jiji reporting carried by Nippon.com. Investigators say the teenager leaned on AI-assisted code generation to build the attack, and the case is being pieced together from server requests, traffic analysis, account-cancellation records, and IP-change history.
It is a small incident with an outsized lesson. The barrier to writing functional attack code keeps dropping as AI coding assistants improve, which means the pool of people who can cause real disruption is widening well beyond skilled professionals. The underlying failure was still an ordinary application vulnerability, the kind that proper access controls and rate limiting are meant to catch.
In short: Tokyo police arrested a 15-year-old who allegedly used AI-assisted code to exploit a Bandai Channel flaw and cancel about 46,800 user accounts.
What it means for your business: AI coding tools lower the skill needed to turn a known weakness into a working attack, so ordinary application bugs on your customer-facing systems now attract a much larger set of potential attackers.
My take: You do not need a nation-state on your threat list to get burned, you need an exposed flaw and a curious teenager with a chatbot. Make sure anything customer-facing enforces real authorization checks, rate-limits sensitive actions like cancellations and password resets, and logs enough to reconstruct what happened. The fundamentals that stop a bored 15-year-old are the same ones that slow down professionals.
Source: Nippon.com
That is the AI and IT news for July 7, 2026. For the last edition, see our July 6 recap, and browse practical how-tos in our Knowledge Base.