THE FUTURE IS HERE, AND IT’S FIXING YOUR TERRIBLE POWERPOINTS

(META), (GOOGL), (MSFT), (ADBE)

I’m sitting in my kitchen at 6:47 AM, watching some overly caffeinated CEO on CNBC promise that his company’s latest AI breakthrough will “fundamentally transform human productivity forever.” You know the drill — same hype, different haircut.

But then the chyron flashed a number that made me nearly choke on my oatmeal: $3.7 billion in AI startup funding. In April alone. 

Now, I’ve seen enough Silicon Valley math to know when someone’s playing fast and loose with their calculator, but this wasn’t venture capital creative accounting. We’re talking about roughly $123 million changing hands every single day, including weekends when most of these founders should theoretically be taking a break from “disrupting everything.”

Here’s what’s ironic about this feeding frenzy: while everyone’s losing their minds over dollar signs, the really smart money is quietly making a much more boring — and profitable — bet.

The AI revolution isn’t happening in some dystopian robot uprising scenario. It’s happening in Susan from marketing finally being able to create a decent product demo video without calling her nephew who “knows computers.”

The industry calls this shift “rich media generation,” which sounds like something a McKinsey intern cooked up while nursing a $14 almond milk latte. But stick with me here.

What it actually means is that AI has stopped trying to pass the Turing test and started trying to pass the “will my boss actually pay for this?” test. Spoiler alert: the second test is way harder.

Take OpenAI’s recent product launches. They’ve got GPT-4o, then 4.1, but they’re phasing out 4.5? Meanwhile, they’re quietly building image generation directly into ChatGPT because they figured out that people would rather create a meme than debate philosophy with a chatbot. Revolutionary? Maybe not. Profitable? Absolutely.

Meta’s (META) playing an even more interesting game, though they’re catching flak for allegedly juicing their benchmark scores — because apparently even AI companies can’t resist the urge to massage their performance metrics.

But while everyone’s arguing about test scores, Meta’s building multimodal AI systems that can actually handle the messy, multimedia reality of how businesses really work. Their new Llama 4 lineup spans from the absolutely massive “Behemoth” model down to the more practical “Scout” version.

Google’s (GOOGL) counter-punch with Gemini 2.5 feels almost defensive, which is fascinating to watch. When was the last time Google looked like they were playing catch-up in anything?

But their focus on complex reasoning tasks suggests they’re betting that enterprises will pay premium prices for AI that can actually think through multi-step problems rather than just generate prettier pictures.

Now, if you’re wondering where to put your money in this beautiful chaos, let me save you some research time. 

The obvious plays are the usual suspects — Meta, Google, Microsoft (MSFT) — but the really interesting action is happening in the companies that figured out how to be Switzerland in an AI world war.

Adobe (ADBE) is quietly becoming the arms dealer of the creative AI revolution. They’re not picking sides in the AI Hunger Games — they’re selling popcorn, tickets, and ad space in the arena. They’re building Firefly as a platform that lets users mix and match AI models like they’re building a playlist. It’s brilliant, really.

While everyone else fights over whose model is superior, Adobe just shrugs and says, “Why not try them all?” Their Creative Cloud integration means they’re essentially renting out digital real estate to AI companies while collecting subscription fees from the users.

But here’s where the tea leaves get really interesting, and why I think most people are reading this market completely wrong.

Safe Superintelligence just raised $2 billion with essentially no product, no revenue, and a business plan that basically amounts to “trust us, we’ll figure out artificial general intelligence.” That’s not a red flag — that’s a Cirque du Soleil performance of financial delusion.

Yet the money keeps flowing because everyone’s terrified of missing the next Google. The problem is, for every Google, there are approximately 47 companies that burned through billions of dollars perfecting solutions to problems nobody actually had.

The companies getting smart money aren’t the ones promising to solve consciousness; they’re the ones solving Excel.

Runway raised $307 million to make video generation actually useful for businesses. Capsule is focusing on helping companies create on-brand content without hiring agencies.

ByteDance is building AI tools that integrate with workflows people already use. These aren’t going to change the fundamental nature of human existence, but they might actually turn a profit.

The European Commission’s new AI Continent Action Plan isn’t just bureaucratic theater either. When politicians start throwing around terms like “digital sovereignty” and proposing to fund 13 AI factories, that’s usually code for “we’re about to start writing very large checks.”

Government contracts have a magical way of turning experimental technology into essential infrastructure.

For those of us actually trying to make money rather than just sound smart at dinner parties, this creates a fascinating investment landscape. The safe play remains betting on companies that have both the resources to acquire promising startups and the existing business models to integrate AI profitably.

But the real alpha (and the real risk) lies in identifying which emerging companies will become acquisition targets versus which ones will join the growing graveyard of overfunded AI startups.

So, here’s my read on the current moment: we’re transitioning from the “AI can do amazing things in controlled demos” phase to the “AI can do useful things in messy reality” phase.

Because sometimes, the future isn’t built on quantum dreams or digital prophets — it’s built on helping Susan from marketing finally make a decent PowerPoint.