🚀 TeachingTechs Tuesday
Your weekly edge in AI, technology, and innovation starts here.
November 10, 2026 | This Week in Tech
Welcome back to another edition of TeachingTechs!
The world of artificial intelligence isn't slowing down, and this week delivered some fascinating developments.
From Nvidia opening an espionage probe into CEO Jensen Huang's relationship with a suspected Chinese intelligence asset OpenAI unveiling a specialized cybersecurity model that wrote nearly a thousand Linux exploits to Anthropic facing a surprise post-IPO "junk" credit warning, Thinking Machines dropping a new #1 open-weights model, and the U.S. midterms triggering a nationwide voter revolt against AI data centres, the future of technology is unfolding right in front of us.
We've rounded up five of the week's biggest AI and tech stories, breaking down what happened, why it matters, and what you should take away.
Let's dive in.
NVIDIA INVESTIGATION
🚨01. Nvidia Opens Espionage Probe Into CEO Jensen Huang Over Alleged Chinese Spy Affair
The world's most valuable chipmaker just lost $190 billion in market value after a bombshell national security disclosure.
On Sunday (Nov 8), Nvidia's board of directors announced an independent investigation into whether CEO Jensen Huang was the "target of a foreign intelligence effort" after he admitted to a past personal affair with Alice Huang, an executive at Southeast Asian cloud firm Megaspeed whom U.S. intelligence officials suspect of funneling information to Beijing.
The controversy escalated rapidly over the past 48 hours:
The Admission: Jensen Huang issued a public apology taking "full responsibility" for his actions while pledging to cooperate fully with the board's national security review and remain on as CEO.
The $4.8B Chip Channel: Megaspeed bought at least $4.8 billion of Nvidia AI hardware between 2024 and 2026 to rent computing power to Chinese giants like Alibaba, and is now under a Justice Department probe over potential illegal chip diversion.
The Market Fallout: Nvidia shares dropped 4.3% on Monday, wiping out roughly $190 billion in market valueand triggering an immediate shareholder lawsuit.
↳ Why It Matters
The global AI race isn't just fought in code or server racks—it's a high-stakes geopolitical battle over hardware supply chains.
When a single hardware supplier sits at the center of the entire AI economy, any governance or national security shock ripples through every tech company on earth:
Tighter Export Crackdowns: Expect Washington to close the "regional cloud loophole" that allows Southeast Asian data centers to rent restricted Nvidia GPUs to Chinese firms.
Key-Man Risk: The $190B stock drop proves just how heavily the AI market's valuation relies on Jensen Huang's personal leadership and government relationships.
Supply Chain Scrutiny: Hardware procurement and cloud partners will face unprecedented counterintelligence and compliance checks going into 2027.
Our takeaway: Even the undisputed king of AI hardware isn't immune to geopolitical risk—reminding every builder why diversifying across chips and model providers is critical.
MIDTERM TECH SHOCK
🗳️ 02. The 2026 Midterms Deliver a Massive Reality Check for Tech & AI Data Centers
Last week's election results just reshaped the rules for how—and where—the next wave of AI gets built.
The 2026 U.S. midterm elections delivered a major political shakeup for the technology sector. While Democrats flipped control of Congress and won key governor races—setting up a divided government against the Trump administration—the biggest shock for Silicon Valley happened on local ballots across the country.
A nationwide voter revolt against AI data center power and water consumption erupted at the polls:
Out of 47 decided local ballot measures across states like Ohio, Kansas, and Arkansas, 46 (98%) voted to ban, cap, or restrict new AI data centers (often capping facilities at 25 megawatts).
Speaking at the WSJ Tech conference on November 5, Meta President Dina Powell McCormick warned that starting new U.S. data center sites is now "very challenging" and revealed that every hyperscaler is now considering building data centers outside the United States.
Even pro-AI super PACs struggled to overcome voter frustration over rising electricity bills and grid strain.
↳ Why It Matters
The election results change the game for the tech industry on three major fronts:
Congressional Subpoenas & Oversight: With Democrats controlling congressional committees, Big Tech, AI labs, crypto firms, and companies tied to federal contracts face a new wave of hearings over AI safety, pricing, and energy costs.
The "Affordability" Lens: Lawmakers in both parties are under pressure to lower household costs, putting corporate AI pricing, subscription fees, and power-grid subsidies directly in the political crosshairs.
A Compute Bottleneck: Because 98% of local ballot initiatives blocked new data centers, existing permitted server capacity just became way more valuable—and new U.S. cloud capacity will take much longer to come online.
↳ How Smart Builders Navigate the Post-Midterm Tech Landscape
If you run an online business or build with AI:
Expect API & cloud prices to fluctuate as hyperscalers face tougher U.S. permitting and look overseas for power.
Adopt lean, efficient models (like MoE or local Edge AI) so your margins aren't squeezed by data center bottlenecks.
Avoid single-platform dependency by keeping your workflows portable across OpenAI, Anthropic, Google, and open-weights providers.
Lead with clear ROI and cost-savings when selling AI products or services to clients in an affordability-focused economy.
Our takeaway: Voters and lawmakers just put a physical and regulatory speed limit on Big Tech's infrastructure binge—meaning software efficiency and multi-model agility are now your biggest competitive advantages.
ANTHROPIC $800B DEBT WARNING
🏦 03. Blackstone Warns Anthropic Could Face a "Junk" Credit Rating After Its IPO
Even a $100 billion IPO might not be enough to make Wall Street comfortable with AI's massive compute bills.
Following Anthropic's blockbuster S-1 IPO filing, private equity giant Blackstone informed investors in a $63 billion chip-financing deal on November 6 that Anthropic is expected to receive a below-investment-grade ("junk") credit rating from major agencies after going public.
That marks a sharp reversal from Blackstone's earlier forecast that the Claude maker would secure an investment-grade rating by the end of 2026.
Why is this such a big deal?
Even though Anthropic holds over $112 billion in cash and equivalents and expects to raise roughly $100 billion in its IPO, it has committed to spending more than $800 billion on computing infrastructure through 2045. Credit agencies look at fixed multi-year debt obligations—and a "junk" (BB-plus) rating means higher borrowing costs and fewer institutional lenders willing to finance those data centers.
↳ Why It Matters
The AI boom is colliding with the hard math of corporate debt markets.
Equity investors love explosive revenue growth, but bond and credit markets care about one thing: predictable cash flow to pay off massive fixed infrastructure bills.
For digital entrepreneurs and solo founders, this is the ultimate validation of staying asset-light. While frontier labs take on hundreds of billions in high-yield debt to build the "railroads" of AI, you get to rent access to their best models via APIs for pennies on the dollar.
Our takeaway: Let the tech giants carry the $800B infrastructure debt—your advantage is building high-margin, cash-flowing businesses on top of their subsidized compute.
OPENAI CYBERSECURITY AI
🛡️ 04. OpenAI’s New "GPT-6 Cyber" Model Found 948 Linux Kernel Exploits
OpenAI just built an AI model that can both break into core internet systems and propose how to fix them.
On November 5, OpenAI unveiled GPT-6 Cyber, a specialized frontier AI model trained specifically to hunt down software vulnerabilities, develop working exploits, and suggest security patches.
In targeted tests on some of the world's most critical open-source software, the model delivered jaw-dropping results:
Discovered 81 previously unknown critical vulnerabilities in the Mozilla Firefox browser (which Mozilla has already patched).
Reproduced 1,724 vulnerabilities inside the Linux kernel and generated working exploits for 948 of them—more than five times as many as its predecessor.
Achieved a leading 48.5% score on the ExploitGym benchmark.
Because of its dual-use power, OpenAI has rated the model "Critical" and is restricting access to vetted security teams and open-source maintainers through its Daybreak program.
↳ Why It Matters
Cybersecurity is entering an automated, machine-speed arms race.
When an AI model can surface hundreds of zero-day bugs and write working exploits on demand, the bottleneck in software security shifts overnight:
Discovery is no longer the limit: AI can find bugs faster than human researchers ever could.
Triage is the new bottleneck: Open-source maintainers and engineering teams now have to verify and patch vulnerabilities at 10x speed.
Shrinking patch windows: The gap between a bug being discovered and weaponized is compressing to near zero.
Our takeaway: If you run websites, apps, or digital infrastructure, automated patching and AI-assisted security auditing are no longer optional—they are survival tools.
OPEN-WEIGHT BREAKTHROUGH
🧠 05. Thinking Machines Releases "Inkling 2 Small" to Dominate Open-Weights AI
A new open-weights contender just matched Google's frontier Flash models at a fraction of the cost.
On November 5, Thinking Machines Lab unveiled Inkling 2 Small, a general-purpose multimodal reasoning model that immediately became the highest-scoring open-weights model outside of China on the Artificial Analysis Intelligence Index.
Using a sparse Mixture-of-Experts (MoE) architecture, the model delivers serious efficiency:
594 billion total parameters, with only 17 billion active parameters used per token.
Full support for text and image inputs with a massive 1-million-token context window.
Matches Google DeepMind's Gemini 3.8 Flash (high) on intelligence benchmarks while costing just $1.18 per complex task (compared to $7.77 for Gemini 3.8 Flash and $10.02 for Claude Sonnet 5.5).
API pricing starts at just $0.30 per 1M input tokens and $1.00 per 1M output tokens, with full Apache 2.0 open weights promised next week.
↳ Why It Matters
High-level AI reasoning and agentic tool use are becoming radically cheaper and more open:
Cost Collapse: Running complex, multi-step AI agents that generate 150,000+ reasoning tokens per task is now 6x to 8x cheaper than using top closed models.
No Vendor Lock-In: With Apache 2.0 open weights on the way, developers and businesses can fine-tune and host a frontier-class multimodal model on their own terms.
Agentic Power: Built specifically for coding assistants, tool use, and RAG workflows, it gives lean teams enterprise-grade intelligence on a bootstrap budget.
Our takeaway: You no longer have to pay top-tier closed API prices to get frontier reasoning—open-weights Mixture-of-Experts models are closing the gap fast.