More than $40 billion is large enough to stop being a collection of deal headlines and become a market structure story. That is the real significance of the latest reporting on Nvidia’s 2026 equity commitments. TechCrunch and CNBC both report that Nvidia has already committed more than $40 billion this year across AI-related equity bets, including major positions tied to OpenAI, Corning, and IREN. On its own, that figure is startling. In context, it is even more important: the dominant supplier of AI infrastructure is no longer just selling into the boom. It is helping finance the boom itself. That changes how investors, founders, and enterprise buyers should read the current cycle. The AI buildout is no longer simply a story of surging end demand pulling hardware into the market. It is increasingly a story in which one of the biggest beneficiaries is also using capital to shape customers, supply chains, capacity expansion, and market confidence at the same time. From chip supplier to ecosystem financier Nvidia has never hidden the fact that it sees investment as strategic. The company’s own investment framing is explicit: it backs “game changers and market makers.” That language matters because it describes a much broader ambition than passive financial investing. It suggests a company trying to shape the conditions around its core business. Earlier reporting already pointed in this direction. TechCrunch’s January review of Nvidia’s startup investing showed a growing pattern of bets across the AI stack, from model companies to infrastructure and tooling. The newer reporting makes the scale harder to dismiss. This is not a side portfolio. It is becoming part of how Nvidia extends its position. There is a logic to that strategy. If Nvidia funds companies building frontier models, data centre capacity, networking infrastructure, or adjacent AI services, those companies are more likely to buy Nvidia hardware, design around Nvidia software, and reinforce Nvidia’s role as the default platform. Capital does not replace product demand here. It can accelerate and organise it. Why this looks different from normal corporate venture investing Corporate venture capital is not new. Large technology companies have been backing startups for decades. But the current Nvidia pattern looks more consequential for three reasons. First, Nvidia sits at the narrowest chokepoint in the AI stack. This is not a diversified platform company making occasional venture bets far from its core. It is the supplier many AI companies depend on to exist at scale. Second, the size of the commitments appears unusually large relative to the strategic centrality of the underlying market. A company that already dominates the picks-and-shovels layer is now also taking a larger role in financing the miners. Third, these bets land during a period when the AI market is still trying to discover what demand is durable, what margins are real, and which business models can stand without subsidy. In that environment, strategic capital can do more than fund growth. It can shape the appearance of growth. The closed-loop risk: when the seller also funds the buyers The sharpest interpretation of this trend is not that Nvidia is doing something improper. It is that the AI economy is beginning to look self-financing. In a normal boom, capital flows from investors into companies that then buy infrastructure from suppliers. In a more circular version, a winning supplier can recycle some of its own gains back into the ecosystem, helping create or accelerate the demand that sustains its lead. That does not make the demand fake. But it does make the signal noisier. This is the key analytical issue. If the same company helps fund the customers, capacity, and surrounding market narrative, then it becomes harder to separate organic demand from strategically amplified demand. Price discovery gets messier. Capital allocation can become less neutral. Public markets may award higher valuations to businesses whose growth is supported, directly or indirectly, by the winners of the previous leg of the boom. That does not mean every Nvidia-linked deal is circular in a simplistic sense. Some are broad infrastructure bets. Some may be financially sound on their own. Some may close on different terms than currently reported. Still, the broader pattern is hard to ignore: capital is no longer just chasing AI growth. Increasingly, AI growth is also being financed by AI’s incumbents. The bullish case: strategic capital can be rational, not manipulative There is a serious counterargument here, and it should not be waved away. The bullish view is that Nvidia is doing exactly what a rational market leader should do in a supply-constrained, winner-take-most environment. If AI demand is real and still early, then helping expand the ecosystem is not distortion. It is execution. From that angle, funding model labs, infrastructure operators, and adjacent suppliers is simply a faster way to remove bottlenecks. It helps create customers that can afford to scale, partners that can support deployment, and capacity that keeps the broader AI buildout moving. In this telling, Nvidia is not manufacturing demand. It is meeting the market where the constraints are. There is also a balance-sheet argument. Nvidia’s cash generation is so large that even aggressive investment activity may still be small relative to its financial strength. If that is true, then claims of bubble dynamics may be overstated. Strategic investing at this scale may look dramatic in headlines while remaining manageable inside Nvidia’s overall economics. That is the best defense of the trend: this may be concentration, not circularity; ecosystem building, not market manipulation. Second-order effects on startups, public markets, and enterprise buyers Even if the bullish case is partly right, the second-order effects still matter. 1. Startups may become more dependent on strategic capital If the biggest checks increasingly come from dominant platform companies, founders may have fewer truly neutral funding options. That can change incentives. A startup may optimise for fitting into a strategic ecosystem rather than building maximum independence. 2. Valuation signals could get blurrier If capital from infrastructure winners helps sustain growth at the application, model, or capacity layer, outside investors may struggle to judge how much momentum is market-driven and how much is subsidy-supported. That can inflate confidence even when the underlying economics remain unsettled. 3. Competitors may copy the model If Nvidia’s approach works, rivals may adopt similar tactics. That would deepen concentration across the stack, with more corporate-backed funding loops connecting chips, cloud, models, and infrastructure providers. 4. Regulators and public investors may pay closer attention As these relationships thicken, scrutiny may follow. Investors will want clearer disclosure around strategic stakes, commercial dependencies, and related-party dynamics. Regulators may start asking whether ecosystem financing can entrench dominance in ways traditional market-share analysis misses. 5. Enterprise buying may get more bundled Enterprise customers may see more infrastructure deals where financing, strategic partnership, hardware supply, and preferred vendor arrangements start blending together. That could speed adoption, but it could also narrow optionality. What to watch in the next quarter A lot depends on whether this year’s headline number becomes a durable pattern. Three things are worth watching closely. First, disclosure quality. Do future filings or executive comments provide more precision on what has actually closed, what remains committed, and how those positions are structured? Second, partner behaviour. Do Nvidia-backed companies deepen commercial dependence on Nvidia hardware and software, or do they maintain meaningful flexibility across the stack? Third, copycat behaviour. If other major AI winners begin using capital the same way, that will suggest the market is evolving toward a more structurally self-financing model rather than a one-off burst of opportunism. Conclusion Nvidia is still the central supplier in the AI buildout. That has not changed. What may be changing is the nature of the boom around it. When the leading seller of infrastructure also becomes one of the most important financiers of the surrounding market, the cycle starts to look different. Less like a pure demand surge. More like a reinforcing loop in which capital, capacity, and commercial dependence feed one another. That does not prove the boom is artificial. It does suggest the boom is becoming self-financing. For founders, investors, and operators, that is the more interesting question now. Not whether Nvidia is winning. It plainly is. The harder question is whether the AI economy is starting to be financed by its winners, and what that does to competition, valuation, and risk. References TechCrunch, “Jensen Huang says Nvidia is pulling back from OpenAI and Anthropic, but his explanation raises more questions than it answers” — https://techcrunch.com/2026/03/04/jensen-huang-says-nvidia-is-pulling-back-from-openai-and-anthropic-but-his-explanation-raises-more-questions-than-it-answers/

Read more →

There are many ways to fuck up when launching your startup, and hiring the wrong people is the most common one. Hiring is always a challenge in business, but large corporations have the resources to absorb a bad hire. For startups, however, a bad hire can be crippling. The high failure rate of startups (around 9 out of 10) tells the story. If you want to be part of the successful 10%, understanding why every hire is critical in a startup and what kind of profiles you should be looking for is crucial. Critical Hire I’m sure you’re familiar with the term ‘critical hire.’ In corporate speak, it’s that fancy title with outsized influence – the VP of Growth God or the Director of Disruption (whatever that means). Jokes aside, generally this refers to a role with strategic importance that directly impacts a company’s direction in a specific area, significantly improves KPIs, and is difficult to replace due to the person’s skills, experience, network, or a combination of all three. I’m sure you’re already starting to see how this definition applies to everyone you hire for your startup. That’s because every hire is essentially a ‘Head of Freakin’ Everything’ until you reach at least 50 people. Why? Because in the startup trenches, it’s all hands on deck. In your early days, you need full steam to bring your idea to life, and only people who understand and share the early-startup mentality can make it happen. There are problems and resolutions that need to be delivered immediately – literally, now. Someone might say this describes a very stressful environment, and that means you’ve grasped the point. It is indeed stressful, but it’s also the perfect environment for people with specific life values and mindsets. You need to be very clear about your expectations when you’re hiring. Now lets talk about why, who and how. Why? You need a very strong understanding of why you need someone, the strategic areas they must cover, and the daily tasks they’ll be responsible for. Communicate this crystal clear (we’ll get to the “how” later). For example, if you’re hiring an engineer to build your platform, you need to understand your customers, acquisition channels, and your primary platform (native app, web, desktop). Don’t worry if you lack technical expertise. Unleash your inner curious George during interviews and you will learn what you need. As a side note, my main degree is in Law, believe it on not. But building my first tech business forced me to learn about microcontrollers, semiconductors, and more through interviews and supplier calls. Often, initial calls initial calls were like staring down a black hole of technical jargon and I didn’t even know what I actually needed. However, by the next call with another supplier, I had enough knowledge for a basic conversation. By the time I finished market research, I could even educate their salespeople (fast learning is crucial). The key takeaway? Listen actively. Talk less, ask insightful questions, and treat interviews as workshops. Moving forward, ask yourself if you can cover the needed business area without hiring. Is it a one-time task, a test project, or a core function? As a rule of thumb, don’t commit to a long-term contract to satisfy some fleeting fancy. If you need someone permanent, invest time in crafting a strong job description. Next, let’s discuss the ‘Who’ section.” Who? As Geoff Smart said in his “Who” book – “Who, is where the magic begins, or where the troubles start”. As I already said, when you are tiny startup, every hire is critical, and Who is the most important part. To illuminate the noise, ideally you want to focus on ex-founders, plan B is to look for people that has a strong sense of life in early stage startups, it can be someone who worked in a few very early startups already for a while and prefect, this is important, such environment. During the interview you want to look for “I like fast passed environment”, “My ideal life-work balance is to work 24/7” (trust me there are plenty of people enjoying such pace, check the Wall street statistics), “Company needs are the highest priority, no matter when”, “get shit done”, “looking for a full ownership”, etc. I was doing a course with 17h long days and 7 days a week for a month just to push myself to the limit, so I know what I’m talking about and I loved it. In the same time having such people requires skills from you. First, is to be able to trust and delegate, any signs of micromanagement and they will run away faster than cheetah. You need to treat them as partners, you need to be fully transparent about every business aspect (marketing, sales, product etc). I always say that do not treat your team members as employees if you don’t want “employees” attitude. Startup life is full of uncertainty, stress and work hard to chase the big dream so you need people that enjoy and thrive in such environment. As usual you must be very clear about that during the interview, you don’t want to waste your time and look for another person very soon, as it might kill your “baby”. To cut through the noise, ideally focus on ex-founders, but those are rare breeds. Plan B? Seek out for people with a strong understanding of early-stage startup life. This could be someone with experience working in several early-stage startups for an extended period, preferably in a similar environment (this is important). Here’s the Vasile’s interview bingo card: “I thrive in a fast-paced environment” (translation: caffeine is my fuel). “My ideal work-life balance leans towards working 24/7” (yes, these people exist – see Wall Street). “The company’s needs are my top priority, no matter the time” (because deadlines don’t sleep). “I’m a get-it-done kind of person” (we need warriors, not wusses). “I’m looking for a role with full ownership” (because babysitting ain’t in the playbook). Full disclosure: I did a course with 17-hour days and 7-day weeks once (talk about pushing your limits!). Let me tell you, I loved it. But that kind of hustle ain’t for everyone. However, managing such beasts requires specific skills from you. Micromanagement is the kiss of death – they’ll bolt faster than a cheetah on Red Bull. Treat them like partners, with total transparency across the board (marketing, sales, product, etc.). I always say: Employees give you employee mentality, partners bring the fire. Startup life is inherently uncertain, stressful, and requires hard work in pursuit of a big dream. You need people who thrive in the chaos. Be upfront about this in interviews. Don’t waste time on the wrong fit. A bad hire can be the final nail in your startup baby’s coffin. Your ideal profile? Ex-founder Risk-taker: Fear of failure? Leave it at the door. Self-driven: Forget hand-holding. They’re internal combustion engines, fuelled by hustle. Results-obsessed: Forget empty promises. They live and die by the metric gods. Full ownership: Micromanagement is a four-letter word. They crave autonomy and bleed responsibility. Communication ninja (we’ll dissect this later, it’s a big one). Words matter. They gotta articulate a vision that rallies the troops. Oh, and domain skills? Yeah, that’s a no-brainer. You wouldn’t launch a spaceship with a pilot who can’t fly, would you? Next stop: How to Find such people. How It’s up to you how many interviews you want to have, but you want to check: You might guess that we will start with job descriptions. Please don’t “look for a rockstar” unless you are hiring for your band. Job description must contain “You will” with the list of specific types of tasks that the person will be doing, those should be very similar to OKRs as they should be easily measurable, “grow daily active users by 20%” not “be awesome. “You have”, also be a list of measurable skills or knowledge that person can demonstrate. Once you have measurable requirements, you can use scorecards to evaluate all candidates equally to make the right decision. You can’t afford a random guess, it’s not a lottery, and again wrong hire can kill your business. Background with references. Always ask for their previous bosses, and if your position includes managing others, ask for references from people they managed. You want to know both sides of the coin. Your startup startup can’t afford a toxic environment, and each new person will have a huge impact on your culture. Competency – always ask “what, how and tell me more”. Ask “What” the candidate was doing in the last few companies. Here you are looking for tasks that might be similar to what needs to be done in your company. Theoretical skills are great but experience is better. “How” is the next big one as you want to know what is the decision-making process, reasoning and how the person approaches different problems. Do they consider alternatives or jump on the first option, it will also show you the experience and the level of domain knowledge. I already talked about “tell me more” when you want to learn things for yourself, but you need to be curious about details in the background, and ask a lot of follow-up questions. If it’s hard to pull the information, take it as a fail on the communication side. For sourcing, always start with your network. The closer you know the person the better as it will help you to get an honest intel on their background, skills and attitude, which means fewer problems with reference checks. Don’t be shy to DM people on LinkedIn, it costs nothing. Direct applicants from your website are the next option as it means the person has a genuine interest in your company, especially when your company is small and none of Google’s algorithms know about your existence. The last is recruiters, but this is when you are really desperate. Recruiters are useful when you are hiring 20+ people at a time but not when you are building kind of a founding team. This approach is not easily scaled, but this is not your problem for now, and if you do it wrong, and scaling might never be a problem. Look for partners who are natural communicators and you will get great results very soon.

Read more →

For the last decade, we’ve trained artificial intelligence to imitate us—our language, our patterns, our preferences. This approach has powered remarkable systems like large language models (LLMs), giving us tools that can write essays, summarise legal texts, and even debug code. But as impressive as these systems are, they’re reaching their limit. That’s the message from David Silver and Richard Sutton—two legends in the field—in their recent paper, Welcome to the Era of Experience. They argue convincingly that if AI is to move beyond mimicry and into true innovation, it needs more than human data. It needs its own experience. This shift isn’t just about new capabilities. It demands a rethink of how we design, monitor, and govern AI. The Core Idea: Learning From Experience, Not Just Data Today’s most advanced models rely on static datasets—Wikipedia, Stack Overflow, Reddit. They’re trained to predict the next word, not to explore or reason in the world. That’s a problem. As the paper points out, “most of the high-quality human data has already been consumed.” We’re near the ceiling of what this approach can offer. In contrast, the next generation of AI will be experiential agents. These aren’t prompt-bound tools. They’re systems that act in the world, observe outcomes, learn from feedback, and adapt over time. They’ll be governed by streams of experience, not isolated queries. Silver and Sutton envision agents that: Retain memory across interactions Pursue long-term goals Learn from real-world signals (not just human ratings) Act independently in digital and physical environments The model isn’t just “chatting.” It’s living, in a way—accumulating experience and knowledge through continuous interaction. The Return of Reinforcement Learning This approach revives reinforcement learning (RL) as the central pillar of AI progress. RL was key to systems like AlphaGo and AlphaZero, which didn’t learn from data—they discovered winning strategies by playing games against themselves millions of times. But RL was sidelined as LLMs took over. The new models were broader, trained on everything at once. But something was lost in the process: the ability to discover new knowledge from the ground up. Now, with experience-based agents, RL is back. And it’s bigger. Agents won’t just explore simulated games. They’ll navigate open-ended environments, guided by reward signals from the world—anything from carbon emissions to heart rate, from exam scores to stock prices. Why This Changes Everything for Compliance From a compliance and governance standpoint, this is the real inflection point—not generative AI, but adaptive AI. Because if a model can: Change its behaviour after deployment Set or influence its own goals Act autonomously across systems and interfaces …then traditional safeguards won’t cut it. We need to answer hard questions: How do we audit models that learn and evolve continuously? What defines misalignment in a system with changing objectives? How do we contain risks when feedback loops span months or years? Reward functions, once hard-coded, may now be dynamic and user-specific. That flexibility is powerful—but it also makes outcomes harder to predict and explain. There Are Safety Gains, Too To their credit, Silver and Sutton don’t ignore the risks. They acknowledge that longer-term autonomy can reduce human oversight and create new failure modes. But they also argue that experience brings unique safety advantages. For example: Experiential agents can adapt to environmental change (like a pandemic or hardware failure). They can adjust their behaviour if users show dissatisfaction or discomfort. They can evolve goals incrementally—avoiding brittle, catastrophic misalignment. In some ways, this is closer to how humans manage risk: through observation, iteration, and course correction. But it still requires a framework of accountability. The Road Ahead: Compliance Must Move With the Technology The era of experience is coming. It’s already arriving in pockets—agents that browse the web, execute code, or simulate long-term planning. These systems won’t stay constrained for long. As AI leaders and compliance professionals, we have a responsibility to stay ahead of this shift. That means: Building standards for continuous oversight Designing reward systems that balance autonomy with human values Developing transparency tools for evolving models Creating protocols for human override—even in persistent learning loops Compliance can’t be static when AI is dynamic. Final Thought The next chapter in AI isn’t about more data. It’s about more learning. Not from us—but from the world. As AI agents grow more autonomous, the real challenge isn’t intelligence. It’s trust, alignment, and governance. We’re not just building smarter systems. We’re building systems that decide how to get smarter on their own. That’s a capability—and a responsibility—we can’t take lightly.

Read more →

Have you heard about the experiment with chimpanzees, poker chips and a vending machine? Let’s dive deep into the 1936 experiment that exposed the OG hustlers: chimpanzees. J.B. Wolfe set up a vending machine – the Silicon Valley of primate motivation – dispensing the holy grail: grapes (or raisins, let’s not get picky). But here’s the twist: these weren’t free. Enter poker chips, the currency of the chimp trade. Chimps, being the natural-born negotiators they are, quickly learned the hustle. Performance for chips? No problem. But then, the scientist decided to switch off the machine and see what would happen. He was surprised to see that they started stockpiling these chips. Forget YOLO (“You Only Live Once”), these chimps were preaching a new gospel – GYODO (“Grapes You Only Desire Once”). Let’s be honest on this point, many people are falling short. As my friend once said when I shared this story, “If my friends were at least half as smart as those chimps, they would be way richer than they are now.” Chimps were demonstrating delayed gratification, a skill long thought to be a human superpower. Imagine this – a chimp, staring down a single grape, muttering, “Nah, gotta play the long game. Multi-grape future, that’s the dream.” This wasn’t just about chimp snacking, it was a paradigm shift. Wolfe’s experiment highlighted a crucial evolutionary leap – the ability to plan for the future. So, the next time you see a squirrel hoarding nuts, don’t judge, they might be onto something. Why people are struggling to save and what you can do about it? We’re gladiators in the arena of modern commerce, bombarded with temptations designed to drain your wallet faster than a retiree at a timeshare sales pitch. I want to highlight three major things: Credit cards and BNPL schemes? Those are the glitter-encrusted handcuffs of debt, promising instant gratification but delivering a lifetime of “minimum” payments. Treat credit like your mortal enemy, and as in any war you need to build as many layers of defence as possible. Every swipe is a mini-battle in the war for your freedom. Ditch plastic cards for non-essential purchases. Go old school – withdraw a set amount of cash each week. When it’s gone, believe it or not, you can’t spend any more (shocking, right?). Feeling the physical weight of your “chips” leaving your hand makes every purchase a conscious decision. And those BNPL apps? Delete your accounts not just apps. They’re the digital equivalent of a wolf in sheep’s clothing. Can’t afford it now? Save up, then pay in full later. Patience isn’t just a virtue, it’s a financial superpower. Now, let’s talk discounts. They exploit our primal fear of missing out (FOMO), triggering a scarcity mentality that screams “BUY NOW OR FOREVER HOLD YOUR PEACE!” The “limited-time offer”, “only a few left” or “50 people are watching now” gimmick is pure retail manipulation, designed to turn even the most disciplined shopper into a mindless clicker. Remember you are not saving 30%, you are spending 70%. Fight back with a personal waiting period. Anything non-essential gets a 24-hour cool-off, and no, new pair of the latest AirPods is not essential. (sorry Tim). Let that initial buying frenzy subside. Ask yourself: “Do I actually need this, or am I just a sucker for a sale?”. The next thing, unsubscribe from marketing emails and turn off sale notifications. Out of sight, out of swipe. Curate your shopping experience. Speaking of curation, social media is another battleground. Are those perfectly curated feeds showcasing everyone else’s #blessed lives? Yeah, that’s a recipe for your financial disaster. The fear of falling behind (thanks, social proof!) leads to impulsive purchases that erode your savings faster than a Kardashian marriage. Remember, until you are really rich, the only person you need to prove something is yourself. Here’s your social media survival guide: unfollow the accounts that make you feel like a financial failure. Fill your feed with frugalistas, minimalists, and financial gurus instead. (Don’t forget to subscribe to my channel). Let them be your digital spirit animals, whispering wisdom and motivation in your ear. Regular social media detoxes are crucial. Unplug for a weekend, a week – whatever you can manage. Reconnect with your real-life priorities. Practice gratitude. Appreciate what you have, not what some influencer is shilling. Focus on building a life that aligns with your values, not the fleeting trends dictated by the latest #OOTD post. Have you watched “The Joneses”? Strongly recommend to watch for understanding how social media impacts your purchase decisions. How to start saving? I’m sure you can outperform the chimp, so your first step is building a habit. Building a savings habit isn’t about clipping coupons like your grandma (although, hey, grandma was onto something). It’s about rewiring your brain for delayed gratification. Think of your future self as your ultimate investor. Every dollar saved today is a future down payment on freedom – freedom from debt, freedom from living paycheck to paycheck. So, 1) ditch the “latte” habit and channel your inner financial samurai. 2) Automate your savings, save at least 20% every month. The goal is 30%+, but we might need to work on your earnings. Don’t worry this is one of my objectives here as well. Set financial goals that make you giddy with excitement (that dream house won’t buy itself!), and track your progress like a hawk. Remember, small, consistent wins are the building blocks of a healthy bank account. Get in the game, and watch your future self high-five you from across the timeline. To summarise So, what’s the takeaway for the modern primate (a.k.a. you)? Saving is not about how much you earn but about discipline, self-control, and habits. To build any habit, you need to create an environment that reduces the risk of losing focus and increases the chances of building it. Learn from your ape ancestors. Master the art of delayed gratification. Don’t settle for the first grape you see. Hold out for the multi-grape extravaganza of life. Remember, in the jungle of success, the chimps are showing us how it’s done. Now go forth and dominate the vending machine of your dreams and prove that Darwin was right about evolution and that we are all better than chimps in every aspect. Start saving now.

Read more →