Image source: Dreamstime<\/em><\/p>\n \u201cHow do you use\u00a0AI?\u201d<\/p>\n That\u2019s my new go-to question when meeting contacts.<\/p>\n Admittedly, it\u2019s self-serving. I get to pick the brains of some of the smartest investors and thinkers in the world about how they\u2019re using the most powerful technology of our lifetimes.<\/p>\n Artificial intelligence (AI) chatbots have transformed my work life. I use Claude, Grok, and ChatGPT at least 50 times daily as my research assistant, writing partner, and instant answer\u00a0machine.<\/p>\n It\u2019s like having a brilliant intern who never sleeps, never complains, and is always ready to\u00a0help.<\/p>\n I\u2019ve even taught my kids how to use AI. Now, when they ask me something I don\u2019t know, they immediately say, \u201cAsk the lady on your phone\u201d\u200a\u2014\u200atheir nickname for ChatGPT\u2019s voice\u00a0mode.<\/p>\n That simple phrase reveals so much about how this technology is becoming woven into our daily\u00a0lives.<\/p>\n Those who say AI is a fad simply haven\u2019t been paying attention. Or they tried the original ChatGPT two years ago, found it mediocre, and haven\u2019t bothered with the quantum leaps made\u00a0since.<\/p>\n It was fine for creating silly pictures or asking ChatGPT to write a limerick about Rhodesian Ridgebacks, but it wasn\u2019t truly useful for serious\u00a0work.<\/p>\n Now? The landscape has completely transformed.<\/p>\n Almost monthly, a cutting-edge AI model debuts that pushes the frontier forward. There are almost too many breakthrough models to keep track\u00a0of.<\/p>\n This rapid progress isn\u2019t slowing down. Look at this graph showing how AI models have been steadily conquering benchmarks, with OpenAI\u2019s o3 and Grok 3 leading the\u00a0pack:<\/p>\n Source: Artificial Analysis<\/em><\/p>\n What\u2019s driving this improvement?<\/p>\n The two raw ingredients you need to make AI better\u00a0are:<\/p>\n Data (books, essays, Wikipedia pages,\u00a0etc.)Compute (Nvidia [NVDA]<\/strong> GPUs processing that\u00a0data)<\/p>\n Scaling laws are brutally simple<\/a>: Crank up the data and computing power by 10X, and the AI\u2019s quality roughly\u00a0doubles.<\/p>\n This is why Elon Musk\u2019s xAI built a monster cluster of 100,000 GPUs, triple the size of the previous record. More GPUs processing more data creates better\u00a0AI.<\/p>\n The new Grok 3 was trained with 3e26 FLOPs of compute. It\u2019s equivalent to running your smartphone continuously for 634,000 years. That astronomical computing power is how Grok 3 smashed benchmarks in record\u00a0time.<\/p>\n Meanwhile, the amount of data used to train top AI models has increased by a mind-numbing 1,018,766,656%<\/strong> since\u00a02010.<\/p>\n It\u2019s hard to wrap your head around \u201c1 billion percent.\u201d It\u2019s like turning $1 into $10\u00a0million.<\/p>\n But that\u2019s the kind of exponential growth AI has achieved in just over a\u00a0decade.<\/p>\n Today\u2019s AI models are pattern recognition systems on steroids.<\/p>\n They identify and learn from text, images, and other data. The more quality data they access, the more they grasp the nuances and complexities of our\u00a0world.<\/p>\n But if your data is trash, your AI is trash\u200a\u2014\u200ano matter how sophisticated your\u00a0model.<\/strong><\/p>\n For example, GPT-2 (launched in 2019) trained on roughly 3 billion words. The entire English Wikipedia, for comparison, contains about 4.6 billion\u00a0words.<\/p>\n Just four years later, GPT-4 trained on approximately 9.75 trillion words<\/strong>\u200a\u2014\u200aover 2,000 times the entire English Wikipedia.<\/p>\n This isn\u2019t some weird academic \u201cwho has the bigger AI\u201d contest. Data is the difference between AI being a cool toy and a transformative tool.<\/p>\n When OpenAI upgraded from GPT-3.5 to GPT-4.5, its score on GPQA (a challenging AI benchmark) jumped from 30% to 70%. More data, better\u00a0AI.<\/p>\n So far, the AI boom has been dominated by infrastructure spending<\/a>.<\/p>\n Tech giants like Amazon (AMZN)<\/strong>, Microsoft (MSFT)<\/strong>, Alphabet (GOOG)<\/strong>, Meta Platforms (META)<\/strong>, and Oracle (ORCL)<\/strong> will spend roughly $300 billion this year on AI infrastructure.<\/p>\n But now, the focus is shifting.<\/p>\n This is a corner of the AI market that\u2019s currently being overlooked.<\/p>\n Right now, everyone\u2019s obsessed with flashy AI models and the chips that power\u00a0them.<\/p>\n They\u2019re missing the critical question: What goes into these machines before they\u2019re even\u00a0built?<\/p>\n Data is the lifeblood of AI. No data, no AI. It\u2019s that\u00a0simple.<\/p>\n Feeding data into AI models becomes much more complicated. It\u2019s not a copy-paste job.<\/p>\n In fact, according to S&P Global, a staggering 60% of an AI developer\u2019s time isn\u2019t spent building cool chatbots or features. It\u2019s spent on the grueling work of gathering, preparing, and cleaning\u00a0data.<\/p>\n It\u2019s imperfect, as data can be reused infinitely and grows rather than depletes.<\/p>\n But in one crucial respect, the comparison is spot-on: Both must be refined before they\u2019re valuable.<\/p>\n When crude oil comes out of the ground, it\u2019s toxic sludge. It must go through complex refinement processes before becoming the gas that powers your\u00a0car.<\/p>\n Data is the same. The raw information companies collect is messy, inconsistent, and often unusable for AI\u00a0systems.<\/p>\n Take a global retailer like Target (TGT). It has reams of customer data\u200a\u2014\u200apurchase histories, browsing patters, demographic information, in-store traffic, etc. But it\u2019s all scattered across hundreds of systems in various\u00a0formats.<\/p>\n Refining it into structured, machine-readable data is where the real money is\u00a0made.<\/p>\n Companies that can clean and structure this data won\u2019t just participate in the AI revolution\u200a\u2014\u200athey\u2019ll lead\u00a0it.<\/p>\n You have to be\u00a0nimble.<\/p>\n The best opportunities usually come from small, overlooked corners of the market. Getting into these opportunities early is how true fortunes are\u00a0made.<\/p>\n Right now, investors only looking at the infrastructure side of the AI megatrend are missing the bigger\u00a0picture.<\/p>\n Don\u2019t be one of\u00a0them.<\/p>\n Remember: While everyone else fights for scraps in the crowded AI sector, the firms that can effectively \u201cfeed the AI beast\u201d with quality data will quietly build the foundation of trillion-dollar empires.<\/p>\n Bottom line: Now\u2019s the time to prepare for this shift\u200a\u2014\u200abefore the crowd takes notice. AI is moving fast, and the biggest winners won\u2019t always be the most obvious\u00a0plays.<\/strong><\/p>\n If you want to stay ahead of the crowd and learn more about the next wave of AI opportunities before they hit the mainstream, <\/strong>sign up for my investing letter The\u00a0Jolt\u26a1<\/em><\/strong><\/a>.<\/em><\/strong><\/p>\n \u2014 Stephen McBride, Chief Analyst at <\/em>RiskHedge<\/em><\/a><\/p>\n Investing in AI: We\u2019re entering the next phase<\/a> was originally published in Coinmonks<\/a> on Medium, where people are continuing the conversation by highlighting and responding to this story.<\/p>","protected":false},"excerpt":{"rendered":" AI focus shifting from infrastructure to\u00a0services Image source: Dreamstime \u201cHow do you use\u00a0AI?\u201d That\u2019s my new go-to question when meeting contacts. Admittedly, it\u2019s self-serving. I get to pick the brains of some of the smartest investors and thinkers in the world about how they\u2019re using the most powerful technology of our lifetimes. Artificial intelligence (AI) […]<\/p>\n","protected":false},"author":0,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[],"class_list":["post-60493","post","type-post","status-publish","format-standard","hentry","category-interesting"],"_links":{"self":[{"href":"https:\/\/mycryptomania.com\/index.php?rest_route=\/wp\/v2\/posts\/60493"}],"collection":[{"href":"https:\/\/mycryptomania.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/mycryptomania.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/mycryptomania.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=60493"}],"version-history":[{"count":0,"href":"https:\/\/mycryptomania.com\/index.php?rest_route=\/wp\/v2\/posts\/60493\/revisions"}],"wp:attachment":[{"href":"https:\/\/mycryptomania.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=60493"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/mycryptomania.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=60493"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/mycryptomania.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=60493"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}Two years ago, I barely used\u00a0AI.<\/strong><\/h3>\n
It comes down to something called scaling\u00a0laws.<\/strong><\/h3>\n
More data = better\u00a0AI<\/strong><\/h3>\n
The biggest investment opportunities are no long in infrastructure alone, but in AI services\u200a\u2014\u200acompanies that feed, deploy, and harness these hungry new machines.<\/strong><\/h3>\n
You\u2019ve probably heard the phrase: \u201cData is the new\u00a0oil.\u201d<\/strong><\/h3>\n
After investing in dozens of megatrends over my career, I\u2019ve learned one crucial\u00a0lesson\u2026<\/strong><\/h3>\n