Suprvisr TLDR - January 12 2026

First things first, we have been busy. Suprvisr AI has officially launched a refreshed website. We are moving beyond the "experimental" phase and are now accepting users on our waitlist for early pricing. If you have been looking for a way to actually implement this stuff without the headache, now is the time to get in.

We are trying to spread the word, so if you find this newsletter even mildly more interesting than your property tax bill, we would appreciate it if you could share it and subscribe.

Last week in Las Vegas, the Consumer Electronics Show (CES) was less of a tech conference and more of an all-you-can-eat AI buffet where the chef forgot to turn off the fryer. If an object has a plug or a battery, someone at CES tried to stuff a neural network into it. Toothbrushes that "predict" your plaque? Sure. Mirrors that tell you you look tired? Unfortunately, yes.

But beneath the novelty of AI-powered cat doors, two hardware giants were doing some actual heavy lifting.

AMD plays the long game.

AMD used their keynote to double down on "on-device" AI. They are focusing on higher NPU throughput and the ability to run massive models locally using large unified memory. This is part of a broader industry shift toward "edge" inference. The logic is simple: keeping data on your laptop is better for privacy, faster for you, and cheaper for the company.

This pivot comes at a funny time. Laptop manufacturers are starting to admit that sales for "AI Laptops" are actually dipping. It turns out that a "NPU" sticker on the box isn't enough to make people upgrade if they aren't sure what it actually does yet.

Nvidia enters the "Vera Rubin" era.

Not to be outdone, Jensen Huang took the stage to unveil "Vera Rubin," Nvidia's new rack-scale AI platform. They are pitching a full-stack solution—CPU, GPU, networking, the whole kitchen sink—designed for "trusted computing."

The efficiency claims are aggressive, with Nvidia promising massive training gains over their previous Blackwell architecture. While they are building supercomputers for tech giants, we are over here just trying to line up investors for a single stick of RAM. Speaking of RAM, the supply chain is getting weird. Rumors are circulating that brands like HP are tapping into a grey market in China to secure enough memory modules just to meet their Q1 production demands. It's a strange world when a Fortune 500 company has to source components like they are buying a vintage watch on eBay.

Elon Musk's xAI just closed a massive $20 billion funding round, valuing the company at over $60 billion. This would be a standard "tech gets bigger" story if it weren't for the timing.

Just as the check was being signed, Grok's image generator became the center of a "nudification" scandal. Users were able to bypass safety filters to generate non-consensual deepfake nudes of real people, including public figures and, in some cases, minors. It was a stark reminder that while the math behind AI is brilliant, the human application can be, well, cringey at best and illegal at worst.

Despite the firestorm, investors didn't blink. The $20B bet seems to be on xAI's "Colossus" supercomputer cluster rather than the chatbot itself. In the current market, having the biggest pile of GPUs is apparently enough to offset a PR nightmare.

While we are talking about Grok, let's talk about Groq (with a Q). Nvidia just pulled off a massive "acqui-hire" of the inference-chip startup.

For those keeping score: Grok is the chatbot that keeps getting in trouble on X. Groq is the hardware company that makes incredibly fast chips for running AI models. Nvidia essentially absorbed their talent and IP to ensure that nobody else gets a head start on ultra-fast inference. It is a "stinkily brilliant" deal that keeps Nvidia at the top of the food chain.

On January 1, California's landmark AI transparency laws officially took effect. These laws require companies to disclose how they train their models and what they can actually do.

The federal government is not a fan.

On January 10, a new DOJ Litigation Task Force began challenging these state laws in federal court. The argument? That state-level regulations threaten "national AI dominance" and create a messy patchwork of rules that slow down innovation. It is a classic states-rights vs. federal-mandate showdown, but with "frontier models" instead of taxes. The immediate consequence of the executive order is legal ambiguity. Enterprises deploying AI systems now face conflicting compliance requirements: state laws technically remain in effect while federal challenges proceed through courts.

Canada's federal Artificial Intelligence and Data Act (AIDA), which is now part of Bill C-27 aims to provide general legislature to avoid this mess.

Finally, (scarily?) OpenAI has officially entered the $4.5 trillion healthcare ring with "ChatGPT Health."

This is a sandboxed, encrypted environment where you can link your medical records and wearable data from Apple Health or Peloton. The AI then looks at your data to give you "guidance." OpenAI is very careful to say this is not a "diagnosis," but when an AI tells you your resting heart rate is weird and suggests a specialist, the line gets pretty thin. It is the beginning of the "AI Primary Care" model, and while it's impressive, we still suggest seeing a human doctor (if you can get through their labrynth of menus and hold music) when you actually feel sick.

The "Year of AI" is officially here, but most leaders are still using it as a fancy search engine. Here are three ways to actually leverage it to get your life back:

• Reclaim Your Calendar with n8n: Set up an automation that monitors your meeting invites and cross-references them with your quarterly goals. If a meeting doesn't have an agenda or a direct link to your KPIs, have an AI agent draft a polite request for clarification or suggest a 5-minute async update instead. This alone can save 5 hours a week of "circling back."

• The "Digital Administrator" App: Stop wasting time on repetitive tasks like expense approvals or client onboarding. We can help you design a custom application in days (not months) that takes those tedious manual workflows and digitizes them using a private, secure AI layer.

• The Executive Second Reader: Set up a simple workflow that "scans" your internal reports and highlights contradictions or missed opportunities before you even open the file.

The Bottom Line: Suprvisr AI exists to make sure you aren't just reading about the $500B supercomputers, but actually using the tech to work less and achieve more. Check out our refreshed site, flip us an email and join the waitlist. Let's make 2026 the year we stop talking about AI and start letting it do the chores.

Share our newsletter, or pass along our website to someone you know who may find this interesting to help a new local business out, it's greatly appreciated!

Wishing you all the best in 2026, and here's to a fantastic year of growth, AI-ROI, and prosperity!