AI Pulse # 2- Jan 7, 2026 - Power Grid Crisis & AMD Takes AI Jobs Stance

12d (edited) • AI Daily

Good morning!

Here's what happened in AI over the last 24 hours:

1. Goldman Sachs Warns: US Power Grids Will Hit Breaking Point by 2030

What happened: Goldman Sachs analysts warned that almost ALL US power grids will run out of spare capacity by 2030 as AI data centers drain electricity at unprecedented rates. Most grids aim for 15% reserve margins. They're already falling short.

Why it matters: This isn't theoretical anymore. The AI boom could literally run out of power. Goldman says this bottleneck could hand China the lead in the AI race if the US can't build infrastructure fast enough.

The numbers: Data center power demand will surge 165% by 2030. US utilities need $50 billion just for new generation capacity. Global grid upgrades could cost $720 billion.

Our take: This is the unsexy infrastructure crisis nobody's talking about. Everyone's focused on who has the best model, but if there's no power to run the data centers, it doesn't matter. Expect energy costs to become a major AI story in 2026.

Source: https://finance.yahoo.com/news/goldman-sachs-warns-us-grids-164159112.html

2. AMD's Lisa Su: AI Is Reshaping Jobs, Not Replacing Them

What happened: At CES, AMD CEO Lisa Su pushed back on the "AI will take all our jobs" narrative, saying AI is changing hiring patterns rather than eliminating roles entirely.

Why it matters: This is the messaging battle of 2026. Tech CEOs need people to embrace AI, but workers are scared. Su's framing it as transformation, not replacement.

Our take: She's half right. AI IS creating new job categories (prompt engineers, AI trainers, model evaluators). But let's be real, some jobs ARE disappearing. The question is whether new ones appear fast enough. Jury's still out.

Source: https://english.cw.com.tw/article/article.action?id=4547

3. Nvidia and Siemens Team Up on "Industrial AI Operating System"

What happened: At CES, Nvidia and Siemens announced a collaboration to create an "Industrial AI OS" for factories. First reference site launches in 2026 at Siemens' facility in Germany. Foxconn, PepsiCo, and others are already testing elements.

Why it matters: This moves AI from chatbots to physical manufacturing. We're talking AI that runs design, production, and operations in real time. This is where the real money gets made.

Our take: This is the unsexy side of AI that doesn't get headlines but will change way more than another chatbot upgrade. If AI can optimize factory floors at scale, that's a productivity revolution.

Source: https://emag.directindustry.com/2026/01/07/ces-2026-ai-nvidia-siemens-samsung-autonomous-industrial-innovation/

4. SEC Sees 700% Increase in AI Related Disclosures

What happened: Companies are mentioning AI in their SEC filings 700% more than they used to, setting the stage for potential litigation as regulators scrutinize AI claims.

Why it matters: Remember when every company slapped "blockchain" on their pitch decks? Same thing's happening with AI. SEC is paying attention now, which means lawsuits are coming for companies making inflated AI claims.

Our take: Good. The AI hype needed a reality check. Expect 2026 to be the year investors and regulators start demanding proof that "AI powered" means something real.

Source: https://news.bloomberglaw.com/legal-exchange-insights-and-commentary/secs-700-increase-in-ai-disclosures-sets-stage-for-litigation

📊 BY THE NUMBERS

165% increase in data center power demand by 2030

$720B estimated cost for global grid upgrades

700% jump in AI mentions in SEC filings

15% reserve margin target that most grids will miss by 2030