• The more organizations invest in AI adoption, the slower their decision-making becomes.
• This dynamic can be described as the innovation–bureaucracy paradox.
• The paradox is simple. The more uncertainty leaders face, the more control mechanisms they introduce. And the more control mechanisms they introduce, the less capable their organizations become of adapting to change.

• Each step feels responsible. Each additional voice feels like protection.

    

     
• Studies describe what is known as the threat-rigidity effect: When leaders face uncertainty or perceived threat, they instinctively centralize decisions, narrow information flows and rely more heavily on established procedures.

• they try to manage the new reality using the control systems designed for the old one.

   

• And once such bureaucratic processes are introduced, they often become the cultural norm: “This is how things are done around here.”

   

• Experienced skiers know the counterintuitive truth: The safest and most effective way to ski down a steep slope is to lean forward.

   
• this reaction often produces exactly what leaders fear most: more confusion, slower decisions, less ownership and ultimately, more bureaucracy.
• Instead of falling back into control, they must lean forward by encouraging ownership amidst uncertainty: clearer decision rights, stronger ownership and cultures that support responsible experimentation.
• They simplify decision processes, reduce approval layers and push ownership to the teams closest to the technology and the customer.

9 more annotations...

• Residents are packing themselves into local county meetings in incredible numbers and calling on their representatives to oppose gargantuan data center projects, developments that could cause electricity prices to spike, drain water supplies, and generate copious amounts of noise.
• Farmers are being hailed as heroes for rejecting millions of dollars to turn their land into data centers,
• at least seven in ten Americans would oppose a data center being built near their home
• 51 percent saying they were against having a data center project near their home,
• “The public has swung 49 points against data centers in just nine months,
• The widespread opposition has turned into a rare bipartisan issue, with conservatives and liberals joining sides to an astonishing degree.
• Younger voters were particularly strong in their opposition, with an overwhelming 83 percent
• Meanwhile, many rural Americans continue their fight while are struggling to have their voices heard

• Worse yet, the region is already facing a water crisis, a dire environmental predicament that could be made far worse by the development.

  Advertisement

7 more annotations...

• At the start of the gen AI boom, the scarcest asset seemed obvious: access to the frontier model.
• Now, beneath all of that, a new constraint is emerging: electricity. The new scarcity is not intelligence but the energy-intensive infrastructure required to produce and deliver it.
• Our research on competitive positioning and optimal distinctiveness and platform globalization and renewable-energy finance suggests that what’s happening today with AI fits a recurring pattern in how firms capture value during technological transitions
• . A new technology creates a scarce control point, and value then pools at that layer, because customers can’t get enough of it.
• The managerial mistake is to keep investing as if yesterday’s scarce layer will endure
• you have to ask what is becoming abundant, what is becoming standardized, and what bottleneck is forming underneath.
• The first era involved infrastructure: the key constraint was connectivity, and firms such as Cisco and AT&T harvested value by controlling the pipes
• The second era involved attention: the key constraint was discovery, and firms such as Google and Meta harvested value by organizing access to information, products, and people
• The third era involved intelligence: the key constraint was compute, and firms such as OpenAI, Anthropic, and Nvidia harvested value from frontier models, chips, and AI infrastructure
• Now the loop is moving into its fourth era: energy & physics. The key constraint in this new era is power, and the bottleneck is shifting away from digital intelligence and back into the physical world: electricity, cooling, land, and grid connections.
• This new era began when AI demand stopped being measured only in parameters, tokens, or cloud budgets and started being measured in megawatts.
• The clearest signal is how aggressively hyperscalers are now moving upstream into generation itself—signing 20-year nuclear-power purchase agreements, acquiring data-center sites adjacent to reactors, issuing RFPs for gigawatts of new power
• Intelligence is getting cheaper but expanding the number of economically viable uses, driving total demand up rather than down.
• Energy is not an input whose price can simply be renegotiated annually. It is local, permitted, slow to build, and politically contested.

• 1) Make energy intensity visible.

   

  Most firms track cloud spend, model accuracy, and AI adoption. Far fewer can answer a more basic question: How much electricity does a given AI workflow require?
• he goal is not perfect measurement. It is to make “intelligence per watt” a management metric, not an invisible engineering variable.

•  2) Reduce demand before buying supply.

   

  Many companies are overpaying for AI because they send too many tasks to models that are larger, faster, and more energy-intensive than the job requires.

• 3) Contract for optionality, not ownership.

   

  Microsoft has signed a 20-year power-purchase agreement with Constellation Energy that is intended to enable the restart of Three Mile Island Unit 1 as the Crane Clean Energy Center, adding roughly 835 megawatts of carbon-free electricity to the grid if the restart is completed.

• 4) Redesign where compute runs.

   

  Cloud-region selection used to be mainly about latency, compliance, and vendor architecture. It’s now also an energy decision. AWS’s acquisition of Talen’s data-center campus adjacent to the Susquehanna nuclear station, for example, and Google’s agreement with Kairos Power for advanced nuclear capacity show that hyperscalers are already organizing AI infrastructure around energy access.

• 5) Make someone accountable.

   

  Create a standing Compute and Energy Council, chaired jointly by the CIO, CFO, procurement leader, and sustainability or operations leader. Some companies are already moving in this direction: Salesforce has embedded sustainability metrics into its AI development process, and Microsoft provides cloud-emissions visibility and sustainability-governance tools that connect technology, finance, procurement, and sustainability decisions.
• As electricity becomes the new constraint, leaders thinking about AI strategy need to stop treating energy as a background input.
• . And they’ll need to recognize that in the next phase of competition, AI strategy and energy strategy will become inseparable.

20 more annotations...

• The document, which outlines Microsoft’s plan to embed a more mainstream and accessible version of OpenClaw AI agents in its Microsoft 365 software suite, describes “three phases” to its approach. The first: “Make people addicted.”

      
• build the skill and tool ecosystem that makes people depend on it daily.
• addiction to me is something no product should be making a part of its build strategy
• “Luckily for us,” they japed, “Microsoft is pretty bad at making addicting products compared to some of the other big companies.”
• Addiction has become one of those scary words in the AI industry.

3 more annotations...

• Many organizations already utilize automation extensively.
• AI agents introduce a different operating model. Rather than following a fixed sequence of instructions, they can evaluate information, make decisi
• The distinction may appear subtle, but the operational implications are substantial. Instead of automating individual steps, organizations gain the ability to orchestrate entire workflows with limited human intervention.

• While artificial intelligence initiatives often begin within information technology departments, many of the most practical applications are appearing within operations functions.

   

• As a result, operational environments provide fertile ground for agent-based technologies.

   
• Organizations that have invested in strong cross-functional collaboration practices may find themselves particularly well-positioned to capitalize on these developments.
• While those benefits are certainly relevant, many operations leaders are finding value in several additional areas.
• Consistency represents one notable advantage
• Organizations also gain improved responsiveness.
• Another benefit involves operational visibility
• The emergence of AI agents does not eliminate the need for operational rigor. In many respects, it increases its importance.

• Before implementing autonomous capabilities, organizations should carefully evaluate process maturity. Key questions include:

   
   
  • Are responsibilities clearly defined?
   
  • Do teams follow standardized procedures?
   
  • Is data sufficiently accurate and accessible?
   
  • Are approval requirements documented and understood?
   
  • Can performance outcomes be measured consistently?
• Most near-term deployments focus on augmenting human capabilities rather than replacing personnel entirely. Employees remain responsible for judgment, relationship management, exception handling, and strategic decision-making.

•  

  Organizations must invest in workforce development, establish clear expectations, and communicate how roles may evolve over time.
• The organizations achieving meaningful results are not attempting to automate entire enterprises overnight. They are identifying targeted operational processes where measurable improvements can be achieved, establishing governance structures, and expanding gradually based on demonstrated outcomes.

13 more annotations...

• Shortly after Uber COO Andrew Macdonald said that it was “getting harder to justify” spending money on AI as it was “very hard to draw a line” from that spend to useful consumer features (after its CTO said Uber burned its entire annual token budget in four months), Axios’ Madison Mills reported that one company had accidentally spent $500 million in the space of a month on Anthropic’s models after failing to set spend limits.
• That’s because, as I’ve said before, nobody can actually measure the ROI of AI, or even create a standard measurement of the cost of a task thanks to the inevitable hallucination-prone nature of LLMs and the ever-growing list of different harnesses and “agentic” (sigh) interfaces
• Yesterday, Microsoft’s GitHub Copilot moved all customers to token-based billing
• Customers are irate. One burned through 50% of their monthly credits in a single prompt, another burned 60% in the space of a few hours, another 31% in a single prompt, another estimated that they’d burn their monthly credits in the space of a single five hour session, another burned nearly half of their credits in eight prompts, another around 14% of their credits in two prompts, and another lamented that GitHub Copilot had gone from their favorite subscription to their most-stressful overnight after burning 33% of their monthly balance in a few hours.
• These users — much like the users of effectively every subsidized AI subscription — never really knew how much anything they did cost,
• Every single user of every single AI subscription service is having their tokens subsidized and the actual cost of AI obfuscated. As a result, every frothy, fluffy hype-piece about Claude Code or AI in general is a kalopsia — the belief that something is more beautiful than it really is. 
• Anthropic, OpenAI and every other AI company deliberately obfuscated these costs because they knew that the second a user actually had to pay for the fuckups of an AI model they’d scream like they were being stung to death by bees.
• Despite Promises To The Contrary, AI Is Getting More Expensive
• CEOs encouraged (or actively incentivized) their workers to use AI as much as possible, in some cases even making one’s AI use a KPI that could cost them their job. 
• “make sure to adopt AI at scale,” all while never, ever having any awareness of what a particular unit of work cost.
• OpenAI, Anthropic and other AI companies have actively conspired to mislead the world about the true costs of AI, and it was working great right up until they decided to try charging what it actually cost.
• Amazon SVP Dave Treadwell told employees to “not use AI just for the sake of using AI.”
• The last few years of AI hype have been built on lies. Every company has conspired to make you think that AI is affordable and sustainable
• Four years and a trillion dollars in, AI is more expensive, its companies more cash-intensive, its products just as unreliable, and its boosters more desperate than ever to make you ignore reality as a means of empowering one of a few ultra-rich oafs
• AI is more expensive today than it was three years ago, and it is not getting cheaper
• . Large Language Models require expensive GPUs thanks to their reliance on power-intensive parallel processing, and larger, more-complex models in turn require more GPUs to both train and run inference with.
• Stop Comparing AI To The Dot Com Bubble — It’s So Much Worse, And Will Not Work Out The Same Way
• The Dot Com Bubble was really two bubbles — one around eCommerce and internet startups, and one around telecommunications infrastructure.
• As a result, infrastructure was built far in excess of what demand existed, because most people weren’t online, and those who were had very slow internet connections.
• Here’s a critical difference between AI and the Dot Com Bubble: when people actually lit up the dark fiber, the underlying internet service was faster, better and cheaper than a dial-up connection
• Dell and CoreWeave just turned on the first Vera Rubin GPUs, and you’ll notice nobody is saying the words “profitable” or “sustainable,” because NVIDIA is not interested in making stuff more efficient rather than more expensive. 
• According to CEO Jensen Huang, AI data centers — which currently cost somewhere in the region of $50 billion per gigawatt — will now cost between $80 billion and $100 billion per gigawatt in the future.
• All of this is to say that the Dot Com Bubble happened due to irrational exuberance and growth lust, and what was recovered at the end came not from scientific breakthroughs but the fact that the useful infrastructure existed and could be adapted and used to make things cheaper and more efficient.

• The AI Bubble Will Not Leave Behind Useful Infrastructure, And Does Not Have A Dot Com Bubble-esque “Redemption Arc”
• An AI data center full of AI GPUs is useful for AI and very little else.
• The entire Dot Com Redemption storyline comes from the idea that it “left behind useful infrastructure,”
• No such story exists for AI.

• The internet and railways didn’t go away because their up front costs were the only real costs that mattered. 
• and the only way to make them even close to generating a profit is to have consistent use of all your GPUs. There’s a cost to having them sit idle — both in electricity and personnel —
• Then there’s the issue of training, which is entirely made up of opex. If you want to train a new model, you’ll likely need thousands — or even tens of thousands — of H100 or H200 GPUs, and they’ll cost just as much electricity whether or not you make anything useful

• Said vagueness created a society-wide gravitational pull of consensus that you needed to be behind AI now, because it’s just like the new internet, except bigger, and if you say it’s not you’re going to be really embarrassed. 

• The only way to inflate the AI bubble both on a hardware and software level was to mislead the general public and investors on the costs and efficacy of AI models. 

• Now that organizations are having to pay the actual cost of AI, suddenly they’re concerned about its outcomes, and everybody has become a little hysterical.

• The Hysteria Around ROI Has Begun

  Nobody Can Measure AI’s Return On Investment Because It Doesn’t Have One

• We are at risk of having an event on the scale of the Industrial Revolution where most of the new output is invisible even as businesses spend increasingly large amounts on AI services.
• SemiAnalysis is a semiconductor analyst firm with an obvious reason to keep the AI bubble inflated,
• We’re four fucking years into it but we’re still using hypotheticals.
• You may be wondering what that “new work done by AI that wasn’t previously being done by humans because AI made it cheap”

• If AI Had ROI, AI Job Loss Would Be Impossible To Ignore

• That’s because AI job loss is a fucking myth. Every company laying off people because of “the power of AI” is doing so because their shareholders are mad and because they know they’ll get headlines. 

• The thing that everybody wants you to avoid thinking about is that if AI worked as advertised, there would be obvious, impossible-to-ignore economic signs:
• The foundation of software would be destroyed, as literally anyone could create and maintain any software they desired.
• The SaaSpocalypse (see my premium here) is a media and market-based hallucination where the collapsing growth of software companies is being explained as “AI taking their business” versus “private equity and venture capital overvalued software companies between 2018 and 2022
• For all of these things to happen, AI would have to be both flawless, hallucination free, a completely different product capable of autonomous intelligence and having unique ideas. 
• The reason that we can’t measure “AI job loss” is because AI can’t do jobs. It can be used to replace some specific contract positions with extremely shitty versions that don’t scale,

• If AI Had ROI, We Wouldn’t Be Discussing Its Potential or Have To Reassure People That It Was “Real”
• Anybody can access ChatGPT, Claude or Gemini, effectively anywhere in the world. The theoretical “power” of AI is that it “just does stuff,” and the proliferation of LLMs would mean that somebody would’ve “done” some “stuff” that we could point at with exceptional ease.

• Whenever Someone Tries To Measure AI’s ROI They Have To Admit It Doesn’t Exist
• Last week, Bain & Co. released a study of 951 executives from companies with more than $100 million in revenue, and unsurprisingly, the data did not declaratively explain what the ROI of AI was:

• Here’s the part that Bain found the most troubling: 44% of large companies that are funding their next wave of AI spending are basing those investments on the last round of savings — savings that haven’t yet materialized for some of them.

• “The technology worked. The value didn’t arrive,” Bain concluded in the report. “Self-funding the next wave from past returns sounds like discipline. In reality, it is a circular bet with a structural leak,” the firm cautioned.

• If We Can’t Measure AI’s ROI, It Does Not Have One 
• This is how the AI bubble inflated! This is how it happened! It happened every time a journalist asked a meaningful question and then immediately diverted to a totally different imaginary topic that made the subject feel good!
• Sam Altman is worth billions of dollars, and OpenAI is allegedly worth $852 billion too, and the best he can give us is “teehee, someone else will work it out,
• If AI had ROI, we wouldn’t be debating whether it had ROI. We wouldn’t discuss its potential, or whether it could, theoretically, under different circumstances, in the future, in a way that nobody can describe be super powerful and do all of the stuff it can’t do today. 

• LLM Users Are The Victims Of A Con
• Is Spotify shipping better features, and are those features launching at a rapid clip? Is the software more secure, or stable? Spotify’s design still looks like absolute dogshit! Most software is worse! Things keep breaking everywhere, and in many cases it’s because of AI coding tools!
• While it might allow some things to go theoretically faster, the overall economic impact of AI-generated code appears to be worse code, worse software, and massive, multi-million dollar bills from Anthropic and Cursor.
• you’re doing more work making the LLM work than you did before they existed, and you’re only proud of it because you feel like you’re special.
• The reason nobody can show you the ROI from AI is that AI does not have a return on investment. Large Language Models can speed up some things in a way that becomes increasingly less-valuable and accurate with the complexity of the task, and more investment in AI data centers does not appear to do anything other than expand the number of tasks that an LLM can attempt. 

• LLMs are sold as “AI,” a technology best-known for automating things, yet they can’t be trusted to run anything on their own. 
• nobody can actually point to the ROI of AI, because it doesn’t exist outside of it burping out stolen content and enriching and ingratiating billionaire dullards.
• You are paying for a tool. You are paying for software. You are a customer. Your job is not to explain to others why this is exciting, nor is it your job to cover up for its mistakes.

61 more annotations...

• 107 milliards d’euros de valeur ajoutée perdue par an et 1,4 million d’emplois non créés,
• ces hausses, qui atteignent en moyenne 8,7 % par an et devraient accélérer à 12 % sur les cinq prochaines années, constituent un transfert massif de richesses hors du continent.
• ces hausses ont longtemps été absorbées en silence, ligne par ligne dans les budgets numériques, sans que leur poids macroéconomique ne soit formellement quantifié.
• Elle ne s’« intéresse plus aux mécanismes de la surinflation, elle en calcule le coût systémique pour l’économie européenne et le traduit en valeur ajoutée détruite, en emplois non créés, en dépenses de R&D et de recrutement sacrifiées.
• Les répondants déclarent une hausse moyenne de 8,7 % par an sur leurs dépenses de cloud et de logiciels au cours des trois dernières années.
• La hausse la plus importante lors d'un renouvellement de contrat atteint 51 % en moyenne, avec des cas extrêmes déclarés à 200 ou 300 % chez certains répondants.
• 'effet prix facial constitue le premier levier (rang moyen de 2,0), devant l'effet volume (3,0), l'effet structure de coût lié à la migration vers les abonnements (3,6), l'effet périmètre alimenté par l'intégration d'outils d'intelligence artificielle (3,6), la baisse de remise (4,6) et le changement de métriques de facturation (4,9)
• Parmi les pratiques commerciales identifiées comme potentiellement abusives lors des renouvellements, le verrouillage fournisseur arrive en tête (40 % des répondants s'en déclarent victimes sur trois ans), devant l'intelligence artificielle imposée par conception (32 %), l'obsolescence programmée (30 %) et la vente groupée (26 %).
• Les 54 directeurs du numérique interrogés anticipent une hausse de 12 % par an sur la période 2026-2030,
• Pas moins de 71 % d'entre eux qualifient cette trajectoire de difficilement ou de totalement insoutenable pour leur organisation.
• 66 % considèrent que le modèle OPEX du cloud représente un risque majeur pour leur capacité d'investissement à long terme.
• La majorité (47 % des répondants) réduit d'autres postes de dépenses numériques, en sacrifiant en priorité les prestations externes auprès des entreprises de services numériques (57 % des organisations qui compriment leur budget IT)
• Un tiers des répondants (33 %) augmente au contraire leur budget numérique global, mais au détriment d'autres enveloppes de l'organisation : la R&D et l’investissement productif (32 % des arbitrages), la trésorerie et les marges (26 %), le recrutement (21 %) et les augmentations de salaire (12 %).
• la modélisation d’Astérès évalue à 65 milliards d’euros par an de dépenses non réalisées auprès d’acteurs européens
• Les 37 milliards de dépenses de prestations aux entreprises de services numériques non réalisées représentent à eux seuls 23 milliards de chiffre d’affaires perdu pour des acteurs européens.
• Les cinq premiers fournisseurs captent à eux seuls 53 % du budget cloud-logiciel des répondants. Microsoft est cité par la quasi-totalité d’entre eux (95 %), suivi de SAP (55 %), Adobe (38 %), Salesforce (35 %) et Snowflake (31 %).
• Les 93 milliards d’euros qui sortiraient définitivement de l’économie européenne représentent la part du surcoût non réinvestie sur le continent,
• Les pertes directes représentent 44 milliards de chiffre d’affaires et 311 000 emplois pour les organisations elles-mêmes
• Les pertes indirectes, chez leurs fournisseurs de rang 1, atteignent 48 milliards de chiffre d’affaires et 318 000 emplois.
• es pertes en cascade dans le reste de l’économie européenne totalisent 113 milliards de chiffre d’affaires et 737 000 emplois.
• e coût annuel moyen des hausses tarifaires en termes de chiffre d’affaires dépasse le budget annuel de l’Union européenne pour 2026 (193 milliards d’euros)
• L’étude rappelle également que le coût des hausses tarifaires équivaut à deux fois l’impact économique du choc pétrolier qu’aurait causé le blocus du détroit d’Ormuz si la hausse du cours du baril s’était maintenue sur un an.
• Les fournisseurs de cloud et de logiciels avancent régulièrement l’intégration d’outils d’intelligence artificielle pour justifier leurs hausses tarifaires
• Les résultats de l’enquête Cigref-Astérès contredisent cette justification dans l’immense majorité des cas
• 52 % déclarent ne pas parvenir à mesurer les gains actuels ou potentiels, soit parce que les effets sont diffus (33 %), soit faute de réorganisation suffisante des modes de travail (19 %).
• La tarification à l’usage de jetons, adoptée ou annoncée par plusieurs acteurs majeurs, constitue un nouveau relais de hausse dont l’ampleur reste difficile à anticiper
• GitHub a annoncé en avril 2026 que son modèle de facturation de Copilot basculerait vers une tarification à la consommation de jetons, une évolution susceptible, selon une analyse du « do tank » Cast x Cigref, de tripler les coûts
• Astérès cite également la décision d’Anthropic de modifier le paramétrage de comptage de jetons sur sa version Opus 4.7, augmentant de 35 % le volume de jetons facturés à périmètre fonctionnel constant
• Sur le plan technique, les charges de calcul, de stockage et de réseau exigées par ces technologies ne peuvent être fournies que par un nombre restreint d’acteurs en position dominante.
• Le premier, dit « levier de productivité », postule que les gains permis par l’intelligence artificielle compenseraient intégralement les surcoûts si ces outils généraient un gain de productivité de 0,8 % par an, ce qui représente deux jours de travail supplémentaires par an et par salarié. Ce seuil correspond à la borne haute des estimations académiques disponibles, qui varient entre moins de 0,1 % (Acemoglu, 2025) et 1,3 % (Aghion et Bunel, 2024).
• Le deuxième scénario, dit « rupture concurrentielle », envisage qu’une pression déflationniste s’exerce sur le marché, soit par la diffusion de l’intelligence artificielle générative capable de répliquer n’importe quel logiciel à coût marginal faible, soit par l’entrée d’un acteur ayant la ferme intention de casser les prix, à la manière de Free dans les télécommunications françaises, qui a entraîné une baisse de 45 % des prix en cinq ans.
• Le troisième scénario, dit « levier de compétitivité », repose sur une politique européenne volontariste de soutien aux fournisseurs locaux. Une hausse de 15 points de leur part de marché permettrait, selon la modélisation d’Astérès, de préserver 120 000 emplois et 9 milliards d’euros de valeur ajoutée par an, à volume de dépenses constant.
• Les dépenses numériques basculent comptablement d’une logique d’investissement vers une logique de coût, transférant de fait l’initiative de l’innovation des organisations vers leurs fournisseurs de cloud et de logiciels.

31 more annotations...

• Leaders are increasingly recognizing that projects are no longer occasional, isolated efforts layered within the traditional organization.
• ow should organizations balance the demands of ongoing operations with the friction of constant transformation?
• What enterprise structures allow high-priority initiatives to move quickly without creating internal chaos?
• In my work, I have studied companies that are meeting these challenges by adopting a new enterprise model: the project-driven organization (PDO).
• I synthesized these inquiries into 10 key questions and answers for HBR readers such as How do you transform without destabilizing the business? and What does this mean for project managers?
• 1. What does a truly project-driven organization look like in practice?

• Most organizations are still built around functional departments designed primarily for operational efficiency—finance, marketing, operations, and so on. Projects typically sit alongside that structure rather than shaping how the organization works.

   

  In a project-driven organization, the balance shifts. Strategic initiatives become the primary way in which talent, resources, and decision-making authority are organized.
• A project-driven organization should not be confused with agile adoption alone. Agile changes how teams deliver work, while a project-driven organization changes how leadership organizes strategy and resources around projects.
• 2. How do you balance operations and transformation without destabilizing the business?

• Leaders often frame this challenge as balancing today’s performance with tomorrow’s growth. Operational work keeps revenue flowing and customers satisfied, while transformation initiatives prepare the organization for the future.

   

  Companies that manage this tension successfully tend to think in terms of initiative portfolios. Some projects improve efficiency within the current business, while others explore emerging opportunities or technologies.

   
• Some organizations address this by deliberately protecting capacity for forward-looking initiatives. Leadership teams identify a limited number of transformation efforts that receive sustained attention even when operational pressures increase.
• 3. How does strategy and planning really change in a project-driven organization?
• here is almost always an evident tension between the need for financial predictability and the reality of market volatility.
• Moving toward a project-driven organization involves shifting from static annual plans to a dynamic prioritization model—one in which leadership teams regularly reassess which initiatives are creating value and reallocate resources accordingly.
• you are essentially moving away from the “illusion of certainty,” that sense of security that comes from knowing exactly where every dollar is allocated in January,
• 4. Why do transformations and projects still fail in capable organizations?
• In many mature organizations, the failure point is rarely “delivery competence.” It is more frequently a structural misalignment across the broader organization—what we might call an ecosystem failure.

• 5. How can organizations reduce hierarchy without losing control?

   
• Reducing hierarchy does not mean eliminating structure. Instead, it often means shifting leadership attention from approving every decision to setting direction and enabling faster decisions closer to the work.
• Many organizations achieve this by defining clearer decision boundaries. Project teams are given autonomy within strategic and financial limits, while leadership focuses on setting priorities and resolving conflicts across initiatives.
• 6. If job titles matter less, how do organizations identify expertise?
• Traditional hierarchies rely heavily on titles to signal expertise and authority. In project-driven environments, however, teams must form quickly around specific challenges, making it more important to understand employees’ capabilities beyond their formal roles.
• Project-driven organizations are experimenting with internal talent marketplaces that make employees’ skills and project experiences visible across the company.
• 7. What is the difference between a working group and a true project team?

• Many initiatives are staffed by individuals who contribute part-time while maintaining their regular responsibilities. While this arrangement allows organizations to involve many stakeholders, it often slows progress.

   

  True project teams typically operate with a higher degree of focus. Members share responsibility for the initiative and devote significant attention to its success.
• 8. What replaces hierarchy in governance and decision-making?

• Traditional steering committees frequently focus on reviewing status updates. While oversight is necessary, this format can slow decision-making when initiatives require rapid adjustments.

   

  Some organizations instead approach project governance more like investment management. Leadership teams review initiatives as a portfolio, periodically assessing performance, reallocating resources, and deciding whether projects should continue, pivot, or stop.
• 9. Why do organizations struggle to sustain transformation momentum?
• Organizations that are trying to accelerate transformation assume that additional projects means faster change. But this is a trap: too many of these companies are running too many initiatives at once, with too little distinction between what is truly strategic and what is simply urgent.
• Employees are spread across multiple projects, often without clear prioritization. Without increasing their capacity to execute, the result is transformation fatigue, not because people resist change, but because they experience constant motion without visible progress.
• In most organizations, fatigue is not caused by too much change, but by too little prioritization
• 10. What does this shift mean for project managers and PMOs?
• As initiatives become central to strategy execution, the roles of project managers and project management offices are evolving. In many organizations, the role is expanding from traditional project management toward broader project leadership—reflecting a shift from managing tasks and timelines to owning outcomes and driving cross-functional alignment.

• Project leaders increasingly act as integrators across the organization. Beyond coordinating schedules and tasks, they align stakeholders, manage competing priorities, and ensure initiatives deliver measurable outcomes.

   
• Project management offices (PMOs) are also shifting from primarily administrative roles toward providing strategic visibility into the organization’s portfolio of initiatives.

33 more annotations...

• Big 4 hyperscalers — Amazon, Alphabet, Microsoft, Meta: about $370B–$410B in 2025, depending on whether you use strict capex, finance leases, and fiscal-year adjustments.
• Adding Oracle, CoreWeave, and xAI/SpaceX AI infrastructure, the practical “AI data-center builder” universe is now around $500B of recent annualized investment and moving toward $700B–$750B+ in 2026 run-rate spending
• With many new companies (Anthropic, OpenAI) going public they will be under pressure to show positive gross margins (Anthropic is close) so they’ll raise prices.
• the high Claude Code costs were already leading them to think about whether they should “outsource” their AI to engineers in India.
• Anthropic has said it is using a new version of a technology called a tokenizer for its latest AI models, which could also contribute to the increased costs paid by customers.
• No matter how you look at it, someone (that’s us) is going to pay twice as much for enterprise software or twice as much for ads (unlikely) for this investment to pay off.
• In other words, all this wonderful AI is quite expensive, and unless it replaces many other things, we’re just going to be paying more
• And from an economic perspective, that means we need productivity, health, or other benefits we have not seen yet.
• Remember, companies like Nvidia, Oracle, Microsoft, and Workday do not plan on “replacing” revenue with AI, they want growth.
• If corporate IT spend goes up (which it clearly is), we need some bigger ROI projects, and willy nilly layoffs aren’t the answer.
• Ill conceived projects will get trashed. In addition to Uber, who burned up their AI budget in weeks, we now find that Pizza Hut and Starbucks had failed AI projects that resulted in $100M lawsuits
• Will the “job elimination” from AI rationalize itself? I think so. We won’t be buying millions of dollars of Claude or Copilot if the cost is higher than doing it by hand.
• Will AI replace other things? Absolutely. Over time the money we spend on other core systems, outsourcing, consulting, accounting, legal services, will all go down.
• Will AI projects become more strategic? Yes, the idea of “giving everyone Claude” may stop, just like “giving everyone a PC” didn’t happen that fast.
• Will AI Agents get smarter? They have to. If that onboarding agent costs $50,000 a month to run, it better be far superior to the call center team that cost $25,000 a month
• Will Consumers stop paying? Hard to say, but if the providers decide that $20 a month isn’t enough to charge this becomes more like a cell phone, where most consumers pay $75-150 per month just for service. And you all know how much resistance there is to that kind of new cost. However if you can start your own business with a $200 per month Claude license, you’ll do it.
• Will massive new markets be needed? For sure. I do believe companies will pay a fortune for models that help with drug discovery, energy engineering, construction engineering, and other “high value” AI value products
• And finally, will the AI “market” adjust? Most likely it will.  The high valuations and unlimited funding for new data centers is highly likely to collapse at some point, and as more people do the kind of analysis I”m doing, more “Warren Buffet” types may opt out of the “buy any stock that’s AI” mindset.

17 more annotations...