Hopefully some of you reading this piece have capitalized on this whole artificial intelligence (AI) arc that we have been witness to the last few years. Let's take a short trip down memory lane to see how this arc began...
Personal computers and infrastructure servers traditionally relied upon a Central Processing Unit (CPU) to run an operating system, applications, logic, and whatever other code that was employed for everyday computing. This chip architecture of a couple powerful cores (individual processing units within a chipset) was great for low latency tasks that were sequential or serial in nature (e.g. fast response for one task.) Along comes a new methodology for building microprocessors referred to as Graphics Processing Unit (GPU.) This chip architecture could be described as "parallel computation" involving thousands of smaller cores that are much less powerful individually but plentiful that gained adoption for graphics rendering, math-heavy workloads, AI training, scientific simulations, crypto mining. The overwhelming parallelism was optimized for high throughput.
When GPUs supplanted CPUs inside servers providing a whole new playground for software, the race for AI started. Unfortunately, I missed the signal that hit me right in the face about 10 years ago. It was when my son in his mid-teens was making serious bank finding GPU cards at retail in one place and selling them elsewhere, generally online. He found that arbitrage opportunity when he began building his own high-powered computer towers to compete in online gaming. He really knew his stuff. He told me I should buy NVIDIA stock. I laughed and did nothing (even at roughly $3/share.) I missed the signal that consumer demand was not being met with current supply of the GPU cards. What was really driving that demand? It turns out - lots of things that we now talk about every day.
NVIDIA stock performance demonstrates how profound this technological shift has been and will continue to be. The race to AI is afoot. Rare earths soon ripped following NVIDIA's explosion in value. Then precious metals (conductors) followed. Looks like power infrastructure buildout is moving the price of copper now. Soon you'll be hearing about power generation - the longest lead time - to finally take us all into the world of the Jetsons. All parts of the compute supply chain have been great investments and will continue to outperform. One should totally exude optimism about what's coming...except the transition is what we are all fearing.
Now that we have all seen what the current form of AI can do - and that is just what is available commercially, not what is in alpha or beta testing - we can see the writing on the wall. Sure, AI can be wrong sometimes, or just outright hallucinate made-up responses to prompts. Those errors will only go down as the models and training tools are tweaked. We will then have trusted access to all knowledge. Profound knowledge. Niche knowledge. Cheap knowledge. It will far outperform humans in this arena both in turn-around time and costs. It's a shame that people who put all this knowledge into books, blogs, articles, documentaries, etc. will not be compensated for training all the AI models. This theft, I predict, will eventually end up at the Supreme Court of the United States.
Who is going to be able to earn money for the knowledge they have accumulated the old fashioned way? AI "commoditized" knowledge and dropped the price to near free. Even at a price point of "near free", the owners of compute capacity saw their wealth explode as labor costs began to vanish. Wage growth collapsed. Human relationships and interpersonal communication have now become a source of friction with a friendly face. Human value is no longer tied to knowledge. Even in the paper industry, we are going to see AI agents monitoring, adjusting, and reporting on real-time conditions. Some entrepreneurial consultants with decades of niche knowledge under their belt are creating their own LLMs that can then be used to earn money solving problems while the consultant sleeps. Who wants to sell hours when you can sell subscriptions?
So what are the pea-brained humans that lack decades of experience and powerful networks of relationships going to do to compete with AI to sell their knowledge? Nothing. You're set up to lose in the entry-level space. However, the glimmer of hope in a world where knowledge is fast and free, is knowing what to ask. I would assert that knowledge was never the bottleneck, or even labor. The bottleneck is curiosity. Famous Silicon Valley investor Naval Ravikant had this to say, "The means of learning are now abundant. It's the desire to learn that is now scarce." Think about that! What will someone be able to do with 10 AI agents working on their behalf? How many billion-dollar companies will launch with 50 employees. Many, I say. Elon Musk put it succinctly, "once you know the question, the answer is usually the easy part."
So the whole AI arc is bringing about a massive change in the character and make-up of the corporate workforce. The end justifying the means is not comforting to any of us. A 2% decline in white-collar employment results in a 3-4% hit in consumer spending. What does that mean when 25-40% of the white-collar workforce is replaced with AI LLMs, algorithms, and AI-based agents? That's why I kicked off this column hoping people made smart investments - they're going to need it through this period of transition.
For those of us making our living in the paper industry, we get to see this play out in a unique position. We actually make things. We are vertically integrated for the most part. We take a natural and sustainable resource, break it down, extrude it in a new form factor, convert the substrate to a useful product (packaging, consumer product) and sell it to the highest bidder. We're in a position to adopt AI to do those things better. We will have access to solutions and case studies that bring new light to our situations. We will be able to analyze data to a degree never before attained in history. Hopefully this results in higher yields, less waste, a greater degree of safety, and a better understanding of the value we generate and how we can position our assets to maximize the value and footprint we control.
Stay curious. Be flexible. Invest while you can. See you on the other side.
Steve Sena (stevesena@me.com) is a Cincinnati native. He obtained degrees in Paper Science & Engineering from Miami University in Oxford, OH and an MBA concentrating in Economics from Xavier University. He's worked for a broad array of leading producers, suppliers, and converters of pulp and paper grades.
