Documenting the headwinds I now see for AI.
It won't seem like it, but I love AI and am long-term positive. But when "math doesn't math" I take note.
1. The core thesis for foundation model lab investment has been high upfront investment made worthwhile by significant long-term profits.
2. These are capital intensive businesses and the compute commitments are very high relative to revenue and require strong growth over long time periods. The "leverage" (commitments versus revenue) is extremely high.
3. The fundamentals are not as positive as they previously were:
• Input costs are higher (commodities, chips, power)
• Interest rates are higher
• Competition is more intense
• Scaling Laws are now problematic: exponential costs/power cannot continue
4. Forecasting compute spend is challenging and high risk due to (a) revenue uncertainty and (b) algorithm uncertainty
5. Revenue growth appears to be slowing. The technology is valuable, but ROI is proving to be more expensive and take longer than anticipated.
6. The future is likely "different models for different use cases" with the lower end of the market being highly competitive.
7. Core use cases such as agentic software engineering are likely to need approaches beyond next-token prediction. They are Σ₂ᴾ complexity problems requiring multi-objective optimization and likely a combination of Transformers and other methods.
8. Current forecasts in memory makers are built largely on quadratic attention. That will not persist: we are already seeing work from DeepSeek, Minimax and Nvidia that can cut RAM needs by 80% or more.
9. This means semiconductor valuations are substantially overinflated and will go through the traditional glut versus shortage cycle.
10. For foundation model providers: lower costs with competitive differentiation is good. However, lower costs with a lack of differentiation would mean lower revenues. This makes it harder to (a) service commitments and (b) pay back investors.
11. Leverage is substantially higher than in previous cycles, evidenced by leveraged ETFs, call option activity and margin loans. Korea is particularly susceptible.
12. 0DTE options create a profile that has stronger parallels to portfolio insurance and 1987 than any other point I can remember.
13. The combination of exponential increases in call activity coupled with the ties of semiconductors to structured products means there is a non-trivial systemic risk to the financial system.
14. Implied earnings growth rates are inconsistent with other periods in history.
15. Macroeconomically we cannot and should not fund exponential cost increases. History has shown us repeatedly that there are better ways (see Quick Sort and Simplex).
16. Significant supply is hitting the market via IPOs.
––
Taken together: costs and competition are increasing while revenue growth is likely slowing. Valuations are fragile and prone to technology disruptions that are already here. Systemic financial market risk is extremely high.
One of the new, buzzy jobs in Silicon Valley is the AI Forward Deployed Engineer (FDE), an engineer who is embedded within a client organization to help customize solutions, such as building and tuning agentic workflows that suit the client’s particular needs. I’ve heard from people who are wondering anew about the FDE career path since OpenAI and Anthropic started building new teams to place FDEs within client organizations.
The rise of FDEs for AI workloads is one way AI is creating new jobs (and why the jobpolcalypse narrative of upcoming job market collapse is false -- there will be many AI and non-AI jobs). However, I believe there will be far more AI Engineer jobs than FDEs, as I explain below.
The FDE role was pioneered about two decades ago by Palantir, which sent engineers to government locations to work on secure, air-gapped networks. In addition to having good technical skills, FDEs need communication skills and sometimes business skills. For example, they may need to speak with clients to understand their needs, formulate a strategy to prioritize projects, explain complex technology, and respectfully push back if a client asks for something unrealistic. They’re enjoying a resurgence because of the amount of work involved in taking an off-the-shelf LLM and building it into a custom agentic workflow that fits particular business needs.
However, I believe the number of AI Engineer jobs will be far larger. A company might accept a few FDEs to be embedded within its organization. But most companies will want far more of their own employees working on their projects. While my organizations do hire FDEs, we hire far more AI Engineers! Also, a common client concern is that it is hard to find vendor-neutral FDEs — they are, after all, there to deeply integrate a particular vendor’s product into a company. In this moment when it’s hard to predict which AI service will be the best one in a year’s time, optionality (the ability to pick whatever vendor turns out to fit best in the future) is very valuable. In contrast, letting FDEs tightly bind a company’s processes significantly reduces optionality.
Right now, I see surging demand for AI Engineers who can build software applications using AI software components (like LLM prompting, agentic frameworks, evals, etc.) and effectively use AI coding agents (like Claude Code, Codex, Antigravity CLI, and OpenCode). As the AI Engineer role matures, I expect it to fragment into more specialized roles, like the generic Software Engineer role from decades ago fragmented into frontend, backend, mobile, data engineering, devops, and so on.
What will be the future, specialized AI engineering roles? I don’t know. Perhaps there will be AI FDEs, LLMOps Engineers, Evals Engineers, AI Data Engineers, Harness Engineers, and other roles we don’t have names for yet. But for now, I see a lot of AI engineers who are generalists create a lot of value. Skilled AI Engineers are in very high demand! As our field continues to mature over the coming decade, I look forward to new specializations within AI Engineering that create even more job opportunities.
[Original text: The Batch newsletter]
🧭 Para las empresas de servicios tecnológicos, el avance de OpenAI y Anthropic hacia servicios de implementación enterprise redefine la oportunidad competitiva.
⚙️ Se potencia el valor de diseñar, integrar y desplegar sistemas de GenAI agénticos conectados a procesos, datos corporativos, controles y operación real.
🛠️ El mercado demandará compañías capaces de construir estas soluciones, gobernar riesgos, controlar costos, asegurar calidad operativa y medir impacto en el negocio.
📈 La diferenciación dependerá de profundidad técnica, conocimiento de industria, cercanía con el cliente y capacidad de traducir modelos fundacionales en resultados sostenibles.
💡 Las empresas de tecnología que desarrollen estas competencias estarán mejor posicionadas para capturar valor en este siguiente ciclo de transformación tecnológica empresarial, centrado en sistemas agénticos y automatización operativa. Para lograrlo, el talento especializado es tan crítico como la estrategia.
👇👇👇
The need and opportunity for professional services and FDEs to deploy agents right now is massive.
Every tech wave offers a new era of consulting and tech services requirements. Moving from analog to digital led to a massive wave in the 90s. Moving from on-prem to cloud did the same in the 2000s. But this is going to be at a scale far greater than the others.
The reason is that agents fundamentally change the underlying workflows of an organization. Unlike most prior eras of technology, where it was a change in medium of the service being delivered (on-prem CRM to cloud CRM), agents rewire the business process itself. And unlike upgrading a tech system, business processes are full of idiosyncrasies.
Every industry will have its own variants, and every department within those industries will have variants as well. Not to mention the bespoke difference between firms. Bringing agents to marketing in CPG will look different from marketing in healthcare. Bringing agents to sales in a B2B software company will look different from a car dealership.
And none of the change is easy technically. You need to first modernize your infrastructure and data and make sure it’s ready for agents; access controls, entitlements, and permissions need to be mapped in a way that works for agents and people; you need to make sure agents have the right context to work with; you need to consistently eval and maintain the agents when there are model upgrades; and you need to drive the change management of the process itself to figure out which parts the people do and what agents do.
That’s an insane amount of technical and domain-specific process work to be done to make this all happen. Huge opportunity for new service providers, as well as internally teams and roles to emerge, to help drive this change.
Trump acaba de anunciar que EEUU bombardeó está noche la Kharg Island, la joya de la corona de la industria petrolera de Irán, 20 km2 dónde sale el 90% de las exportaciones. Ningún otro gran productor de petróleo depende tanto de una sola instalación. La isla representa el 2% del petróleo global
Trump dijo que eligió no destruir la infraestructura petrolera de la isla "por razones de decencia", pero advirtió: si Irán o cualquier otro interfiere con el libre pasaje de barcos por el Estrecho de Ormuz, "reconsideraré inmediatamente esta decisión. Es el próximo paso
El conflicto sigue escalando. Ahora uno entiende la cara de Bessent en la entrevista de hoy. El precio del petróleo camino a los usd 150
Uruguay’s central bank extended its easing cycle, lowering the benchmark interest rate by three quarters of a percentage point to 5.75% as it grapples with inflation below its target https://t.co/n2Wjtr6syd
The future of work will look something like what Boris is describing. Anthropic is hiring engineers because people who know what they’re doing still have to tell the agents what to do, review their work, and integrate that work into a broader system.
This will be true of other types of work as well; we will just move to higher levels of abstraction. It may be hard to imagine how that doesn’t lead to the evaporation of work, but once you consider all the natural limitations of agents it becomes clearer what the roles will look like.
Also, as you automate one part of a process you quickly discover the bottlenecks in another part of the process. Many new forms of work will grow simply because another type of work became more efficient and eventually is constrained somewhere else in the system.
This is how you can square the idea that more and more of today’s tasks can be automated, yet you still end up needing people to manage all those tasks.
Para hablar con números concretos, ahora mismo USD 4542.24 de gastos locales por 4 contenedores.
Estamos hablando de USD 40 por tonelada (eso si lograste meter 28 tons en el contenedor, pues hay países que limitan el peso a 24 tons brutas y ahí el costo por kg se dispara).
Hay productos en los que eso representa el margen.
El puerto hoy, por sus costos e inconsistencias operativas (que a fin de cuentas representa costos) es un freno al desarrollo del país.
Salió la Unión de Vendedores de Nafta a quejarse de que el 25% de los autos 0 km que se venden son eléctricos, y en una semana subieron 5% el costo de carga y eliminaron el subsidio a la compra de autos eléctricos. Y después dicen que el Estado no es eficiente...
Comercio con EEUU - Ahora EL TREN VIENE DE FRENTE.
El Acuerdo entre Argentina y Estados Unidos es un cimbronazo para el escenario regional con consecuencias potencialmente importantes en la economía y en los agronegocios.
https://t.co/K7SgiSjw9x
I'm not taking a stance on whether inflammation drives cancer, but I will say it's very true that GLP-1 drugs reduce inflammation—a lot!
Tirzepatide at any dose greatly reduced levels of high-sensitivity C-reactive protein and interleukin-6, two important inflammation markers:
🚨: So..... it happened!
3I/Atlas did accelerate non-naturally at perihelion to navigate away from the sun. It also unexpectedly increased in brightness.
There are potentially natural, albeit far-fetched, reasons this could happen on its own, but add those low probabilities to the list of already very low probability explanations and either this comet is a combination of all the rarest possibilities, or... it is non-natural, intelligently designed and controlled.
Occam's Razor says it is the latter. The simplest explanation is usually the truest. And when all other explanations are exhausted, the one that is left, no matter how improbable, is likely true.
This satisfies both checks.
Keep your eyes to the sky this November!
NEW - Research finds no evidence that aging or declining populations harm social and economic performance. On average they perform better overall, in wealth, productivity, and life expectancy.
En países como España, donde la natalidad se desploma y la población envejece, la inmigración puede ser una herramienta positiva para revitalizar la economía y sostener el sistema de pensiones. Sin embargo, el verdadero problema no es la inmigración en sí, sino su combinación con un estado de bienestar mal diseñado.
Cuando se permite que cualquier recién llegado acceda de inmediato a beneficios sociales sin haber contribuido previamente, se genera un incentivo perverso: el sistema atrae a quienes buscan aprovecharse, no a quienes quieren aportar.
Una solución sensata sería establecer un periodo mínimo de cotización—por ejemplo, 20 años—antes de tener derecho a prestaciones no contributivas. Así nos aseguraríamos de que quienes inmigran lo hacen con vocación de integrarse, trabajar y sumar al país que los recibe. Y el que deja de cotizar vuelve a casa.
Inmigración, sí. Pero con reglas claras y justas. Un país solidario no puede ser ingenuo.
One open AI Agent UX question for the industry is whether products will have many agents that are deployed to perform individual tasks, or have one agent that performs many tasks at once.
While either approach leads to similar consequences in the core architecture design, they tend toward different mental models of how to interact with agents at scale.
Do you build your software to deploy lots of agents and manage their workflows? Or do you build software to deploy an agent across many workflows? The logging, auditing, orchestrating, and even API designs will differ based on the approaches that are taken here.
And as people, developers, and companies deal with hundreds, thousands, or millions of agents, the paradigm that we all collectively land on will start to matter more and more.
Thousands, and then millions, of American small businesses, including many iconic brands, will go bankrupt this year if the tariff policies on China don’t change.
🧵