FhtAbd

A DEVELOPER STOOD UP AT A CONFERENCE AND APOLOGIZED TO THE ENTIRE INDUSTRY FOR ONE LINE OF CODE HE WROTE IN 1965. THAT SINGLE DECISION HAS SINCE CAUSED CRASHES, BREACHES AND ROUGHLY A BILLION DOLLARS OF DAMAGE AND YOU HIT IT EVERY SINGLE DAY 61 minutes from Tony Hoare -- the man who invented quicksort, won the Turing Award, and passed away in early 2026, leaving this as one of his most honest talks. -> His confession: he added the null reference, letting any variable secretly mean "nothing", simply because it was easy to implement. That one shortcut became the null pointer, the undefined, the None -- the crash waiting inside almost every program ever written. He calls it his billion-dollar mistake, and he is not exaggerating. Whole classes of bugs and security holes trace straight back to it. And this is the deeper lesson under the AI rush: the easy shortcut you ship today becomes the landmine everyone steps on for the next 40 years. You thought null was just part of how code works. This is the talk where the man who made it tells you it never had to be. Bookmark & Watch this one. You'll never type null the same way ↓

Our thoughts on the importance of AI sovereignty. 1. Your AI sovereignty dictates your institution’s future. Sovereignty is the precondition for choice. Relinquishing sovereignty transfers the future choices of your institution to others, who are likely to exploit it for their gain and your loss. 2. Data retention is your treasure. Transfer it at your own peril. Your ability to win is dictated by your ability to recognize and use your unique edges, and you keep winning by compounding the underlying data to generate new insights. Transferring that data hands over access to your pre-existing winning plays and yields the means of production for new ones. 3. Tokenmaxxing hijacks your value orientation and decreases your institutional fortitude and intelligence. The pursuit of high token usage incentivizes disposable scripts over robust software — with the addictive feeling of false progress. There is a reason why those selling tokens refuse to charge based on value. 4. Controlling your weights is controlling your fate. Weights are the distilled form of hard-won, accumulated institutional knowledge. If you let others control your weights, you are allowing them to migrate the alpha of your business to theirs. 5. There is no contradiction between sovereignty and alpha. The architecture that maximally preserves sovereignty is one that enables institutions to own their tribal knowledge, and to compound it as alpha. 6. Politicizing the technical issues involving sovereignty is what your adversary wants. Techno-politicization is the wellspring of false sovereignty. Techno-politicization drives decisions that seem to reduce dependency, but ultimately limit agency — especially on the battlefield in the West. 7. Real expertise is existential. Allowing politics or favoritism to determine your technical decisions rewards whoever is best at politics, not whoever is right. Listen to those closest to the problems, not those speaking most compellingly about them. 8. Learn from institutions that are winning or that have consistently delivered. Institutions facing existential threats do not have the luxury of making technical decisions based on political preferences. 9. Only listen to institutions, countries, and people who have a proven record of being right. A track record of correctness is the best and only signal for future correctness. Judging something as right or wrong based on who you like is exceedingly misguided.

Our thoughts on the importance of AI sovereignty. 1. Your AI sovereignty dictates your institution’s future. Sovereignty is the precondition for choice. Relinquishing sovereignty transfers the future choices of your institution to others, who are likely to exploit it for their gain and your loss. 2. Data retention is your treasure. Transfer it at your own peril. Your ability to win is dictated by your ability to recognize and use your unique edges, and you keep winning by compounding the underlying data to generate new insights. Transferring that data hands over access to your pre-existing winning plays and yields the means of production for new ones. 3. Tokenmaxxing hijacks your value orientation and decreases your institutional fortitude and intelligence. The pursuit of high token usage incentivizes disposable scripts over robust software — with the addictive feeling of false progress. There is a reason why those selling tokens refuse to charge based on value. 4. Controlling your weights is controlling your fate. Weights are the distilled form of hard-won, accumulated institutional knowledge. If you let others control your weights, you are allowing them to migrate the alpha of your business to theirs. 5. There is no contradiction between sovereignty and alpha. The architecture that maximally preserves sovereignty is one that enables institutions to own their tribal knowledge, and to compound it as alpha. 6. Politicizing the technical issues involving sovereignty is what your adversary wants. Techno-politicization is the wellspring of false sovereignty. Techno-politicization drives decisions that seem to reduce dependency, but ultimately limit agency — especially on the battlefield in the West. 7. Real expertise is existential. Allowing politics or favoritism to determine your technical decisions rewards whoever is best at politics, not whoever is right. Listen to those closest to the problems, not those speaking most compellingly about them. 8. Learn from institutions that are winning or that have consistently delivered. Institutions facing existential threats do not have the luxury of making technical decisions based on political preferences. 9. Only listen to institutions, countries, and people who have a proven record of being right. A track record of correctness is the best and only signal for future correctness. Judging something as right or wrong based on who you like is exceedingly misguided.

⚡️Karp is naming the real enterprise AI split. The frontier labs sold intelligence as a universal product. Enterprises are discovering that raw model access is not enough, and in sensitive environments it can become a sovereignty problem. The issue is not only performance. The issue is control. Who owns the model behavior? Who owns the weights? Who owns the prompts? Who owns the data exhaust? Who owns the workflow knowledge? Who owns the company’s operational alpha after the model has observed it? That is the center of the rant. Karp is saying the frontier model layer is trying to become a tax on every enterprise’s private knowledge. Companies pay token fees, hand over context, expose workflows, and then fear the model provider can learn from, replicate, commoditize, or eventually compete with the very business it serves. Even if the lab says it will not do that, the trust problem remains because the customer does not control the full stack. That is why Palantir plus Nvidia matters. The pitch is: keep compute, models, ontology, data, and operational logic under customer control. Use models as replaceable components. Do not let OpenAI, Anthropic, or any single frontier lab become the owner of the enterprise brain. The strongest line is “they want to know they own the means of production.” That is the whole architecture war. In consumer AI, users tolerate dependence. In enterprise, battlefield, manufacturing, healthcare, intelligence, energy, and regulated finance, dependence is unacceptable. The model cannot be a black box that absorbs secret workflows and charges tokens forever. The model has to be embedded inside a controlled operational layer where the customer owns the data, governs the model, controls compute, switches providers, and preserves institutional alpha. This is extremely bullish for Palantir’s ontology thesis. Karp is basically saying the value is not just the model. The value is the application layer that turns models into usable, safe, governed action inside real institutions. That is exactly Palantir’s lane. Frontier labs provide cognition. Nvidia provides compute. Palantir provides the operating structure that makes cognition useful in environments where mistakes, leaks, hallucinations, IP loss, or data exposure are existential. That is also why he is attacking token economics. Token billing works when AI is a tool. It becomes offensive when AI is sold as transformation but the customer cannot see value, cannot control the stack, and suspects the vendor is capturing the business’s hidden knowledge. Enterprises do not want to rent intelligence from a potential future competitor. They want sovereign capability. The AI bubble point is more precise than “AI is fake.” Karp is not saying AI is fake. He is saying the frontier-lab business model may be mispriced, overtrusted, and structurally wrong for serious enterprise deployment. Compute plus application layer plus model is real. Raw model subscription sold as enterprise transformation is weaker. That is the knife. So the likely outcome is not enterprise AI demand collapsing. The likely outcome is enterprise AI spend migrating away from generic chatbot/token subscriptions and toward controlled stacks: private compute, open or controllable models, Nvidia infrastructure, secure application layers, ontology, governance, auditability, and domain-specific deployment. That is bullish for Nvidia. Bullish for Palantir. Bullish for sovereign AI stacks. Bullish for open models when paired with secure deployment. Bearish for frontier labs that assume they can own the customer relationship, the model layer, the data interface, and the economics forever. The deeper geopolitical read: battlefield AI cannot be outsourced to consensus Silicon Valley. Karp is saying American warfighting, critical infrastructure, and industrial command systems need AI, but they cannot be dependent on model providers whose governance, safety ideology, product incentives, and data posture are misaligned with sovereign use. That is why he keeps saying Department of War, Ukraine, Israel, critical infrastructure. He is framing Palantir as the patriotic enterprise operating layer between uncontrolled frontier labs and national power. The cleanest read: The AI stack is splitting. Model layer: powerful but commoditizing and politically distrusted. Compute layer: scarce, profitable, strategic. Ontology/application layer: where enterprise value, trust, control, and workflow ownership live. Karp is arguing that the model companies overplayed their hand by acting like the model owns the future. Enterprise customers are realizing the future belongs to whoever controls the operational layer around the model. That is exactly Palantir’s thesis. And he is probably right.