Pablo Castro

Andrew Ng reported from Davos, that AI deployment for transformative impact will require workflow redesign. It is "theory of constraints" all over again (remember Eli Goldratt's "The Goal"?), we have to understand that real business benefits come only from working across silos, from analyzing flow bottlenecks, from understanding that optimizing non-bottlenecks creates waste. As Ng noted, letting a thousand (AI) flowers bloom has failed to lead to significant payoffs. Obviously! We in the silicon industry are no strangers to this. Workflows have been stagnant for decades. Technologies for high level specification, high level synthesis, and formal verification are not yet integrated into the full silicon development flow. They are often done piece-meal, and enthusiasts' pilots are dropped at the hint of a failure or delay. People say they want formal verification in their flow, but want it yesterday and do not have the time to invest in training engineers for its rigorous usage. Ashish Darbari, responding to a question on "if formal verification is so powerful then how come it is not widely used", noted that the problem is in the name - the actual task to be done is formal analysis while the bug detection is only a corollary, but the industry puts the cart before the horse. Based on my experience with industrial practice of high level specifications and formal verification, I tend to agree with his succinct comment. Jan Rabaey, in a DAC keynote speech, said that someday he expects to see a comic book series entitled "adventures in formal verification". AI deployment faces the same problem, if not actively worked on by business leaders. To be fair, deployments of all disruptive technologies face this mental barrier, we are so used to thinking in terms of the old flow (using the old technology) that thinking of a new flow literally means us having to begin again from first principles - few leaders have the courage and the time to do that. Goldratt mentioned that he himself had faced the same mindset problem when deploying the early ERP solutions and was surprised why he wasn't getting the expected large benefits, and he said it took him many years before he could identify the core problem and define solutions for it, which he named (rather academically) as "theory of constraints". AI leaders would do well to rework the conventional workflows in a holistic manner in order to get the real benefits from this new technology. As Henry Ford had said, "If you always do what you've always done, you'll always get what you've always got".

21

Jensen Huang just called OpenClaw "the most important release of software, probably ever." When it comes from the man who defined the GPU era, then people listen ( after discounting the vested interest obviously :) While people focus on his quote, we should understand that the real story is adoption speed of Openclaw: it took 3 weeks to reach Linux's 30-year milestone. That's not growth, but an explosion. OpenClaw isn't a tool. It's the operating system for agents that do work. When NVIDIA's CEO calls a project THE thing, every CTO adds it to their roadmap. ByteDance, Alibaba, Tencent are integrating OpenClaw —not for GitHub stars, but because they're terrified of falling behind. On March 6, 2026, nearly 1,000 people queued outside Tencent's Shenzhen headquarters for free, on-the-spot installation of OpenClaw AI. That shows the desperation to be in-step with the technology. In many orgs, OpenClaw is already in production. JPMorgan handles fraud detection with it, Walmart wired it into loan approvals. These are real systems with real money and stakes. The Agenci AI market is fragmented: LangGraph, CrewAI, AutoGen, LangChain, Semantic Kernel, smolagents, PydanticAI. Seven frameworks fighting. But the outcome is obvious. Whoever controls the execution layer controls the decade ahead. The builders who shipped agents in 2024 when LLMs were not ready for it, they already know the significance of these events in 2026 . Which side of this are you on?

74

22 Comments

A recurring ambiguity arises around a fundamental architectural question: where do generative AI models fit within the technology stack? Are they infrastructure primitives, akin to databases and message queues, middleware services, or merely application-level integrations? Many organizations have yet to resolve this, and the downstream impact is material. In my view, generative AI models are infrastructure primitives and should be positioned alongside databases and message queues, rather than as application features. This perspective shifts the focus to what truly matters, as the abstraction layers above must now bear the responsibility of translating raw capabilities into business value. When infrastructure is this potent, engineering rigor and product discipline in the layers above become essential; mediocrity can be amplified at scale rather than concealed. A well-considered product decision has a multiplying effect across every deployment, while gaps in architecture or engineering rigor can lead to customer friction and operational challenges. Evidence of this is already visible in organizations attempting generative AI integration without reevaluating their infrastructure assumptions. They are experiencing the costs in real time, with structural misalignment becoming apparent at an organizational scale. Reading through a recent article reaffirmed my belief that generative AI models belong at the infrastructure layer. https://lnkd.in/ggQQhcxW The key question is not whether you are using generative AI, but whether you have thoroughly considered architecture, product discipline, and engineering rigor at every layer above infrastructure, as that is where value or loss ultimately accumulates.

38

2 Comments

Indeed. A small number of big tech companies have grossly overinvested in LLMs and overextended themselves -- even laying off tens of thousands of critically needed human employees and enshittifying their primary products and services to reallocate investment into it -- shooting Nvidia's profits and shares to the moon -- all in an insane attempt to "be first" for something that has no forseeable compelling ROI. It's a collective delusion and hype bubble. This is markedly different from the strategic "operate in the red to build marketshare" strategy of the few survivors of the dot-com bust, because at least those survivors were able to articulate (and realize) a logical ROI after reaching a tipping point of critical mass. This is more like the 98% of dot-com startups that burned bright fueled by VC right until they burned themselves out because they never actually had any believable plan for monetizing anything they were building. The lack of ROI will come rudely crashing down on these companies over the next several quarters as they run out of tricks to cook the books and hide the losses. Shareholders will get spooked and back out, share prices will crash, and big tech companies that have bet the farm on LLMs will suddenly find themselves without the necessary employees, capital, quality products/services, or reputational sway needed to save themselves. And when they collapse, Nvidia will collapse with them. And then maybe, finally, we can get back to the business of making things of actual value to humans.

19

Someone tell the wall street experts. LLMs don't create expertise out of thin air. Undervaluing the companies with the existing knowledge base is a career limiting move.

11

The talent wars at the biggest AI companies keep heating up. Senior people do not switch companies for vanity. They move because the work moves faster and the road to real impact looks clear. Inside these labs, access to large compute decides the pace. People care about authorship credit and a voice in model direction. Titles only matter when they unlock both. Leaders who want to compete must make three clear promises. - Ship decisions on a weekly schedule. - Give top talent a seat at the table for model choices and evaluation rules. - Offer serious compute without ceremony. One hire does not decide a race. A run of hires that concentrates compute, authorship, and product velocity does. Watch for that pattern.

3