Get Your Weekly Digest

Receive the best podcast summaries and insights delivered to your inbox every week.

We respect your privacy. Unsubscribe anytime.

Revisionist History

IBM CEO Arvind Krishna: Creating Smarter Business with AI and Quantum

with Arvind Krishna

Published November 27, 2025
View Show Notes

About This Episode

Host Malcolm Gladwell interviews IBM CEO and chairman Arvind Krishna in front of a live audience at IBM's New York City office about IBM's role in solving complex business problems through technology. Krishna reflects on his early technical career, his predictive bets on networking and streaming, his strategic decision to acquire Red Hat instead of chasing hyperscale cloud, and his views on how enterprises should pragmatically deploy AI. He also explains why he believes quantum computing is a third, fundamentally different form of computation on par with the semiconductor revolution and outlines a near-term timeline for impactful quantum applications.

Topics Covered

Artificial intelligence Quantum computing Leadership and management Systems thinking Entrepreneurship Business growth and scaling Problem-first invention Investing Personal Development Innovation strategy

Disclaimer: We provide independent summaries of podcasts and are not affiliated with or endorsed in any way by any podcast or creator. All podcast names and content are the property of their respective owners. The views and opinions expressed within the podcasts belong solely to the original hosts and guests and do not reflect the views or positions of Summapod.

Quick Takeaways

  • Arvind Krishna defines IBM's mission as improving clients' businesses by deploying the right technologies across hybrid cloud, AI, and emerging quantum computing, rather than pushing specific products.
  • His early work on wireless networking and predictions about the fusion of networking and computing shaped his conviction that technologists must also understand markets, business models, and routes to scale.
  • The much-criticized 2018 acquisition of Red Hat, initially punished by markets, later became widely seen as IBM's most successful software acquisition and a cornerstone of its hybrid cloud strategy.
  • Krishna argues that enterprises are underusing AI in high-impact, scalable areas like customer service and software development productivity, while chasing shiny but low-value experiments.
  • He believes current large language models will not on their own lead to AGI and stresses the need to combine them with explicit representations of knowledge, alongside major gains in compute efficiency.
  • Quantum computing, in his view, introduces a third kind of math that will unlock solutions to previously intractable problems in finance, materials, and optimization within three to five years.
  • Demographic decline in developed countries makes AI essential to maintain living standards by augmenting labor, while in developing countries AI can unlock access to advanced capabilities at low cost.
  • Krishna emphasizes persistent, patient leadership that pushes hard on big bets like quantum while fostering open dissent, broad internal and external advisory networks, and continuous personal learning.

Podcast Notes

Introduction and framing of IBM's role

Malcolm Gladwell introduces the show and season theme

Malcolm states that the podcast is "Smart Talks with IBM" and that this season highlights how IBM helps clients solve complex problems[0:14]
He mentions examples like helping L'Oreal rethink cosmetic formulation and enabling Scuderia Ferrari HP to connect with fans in new ways[0:20]

Introducing Arvind Krishna and the focus of this episode

Malcolm says the episode will zoom out to look at IBM's bigger picture through a conversation with its CEO and chairman, Arvind Krishna[0:34]
He notes they recorded in front of an intimate live audience at IBM's New York City office[0:37]
Key themes he flags: Arvind's ability to anticipate tech directions, the future of AI, and his passion for quantum computing, which Arvind calls as revolutionary as the semiconductor[0:48]

What IBM does and Arvind's definition of its mission

Malcolm's long-standing question about IBM's identity

Malcolm recalls asking his IBM-employed cousins over the years what IBM does and getting confusing, different answers[1:13]
He asks Arvind for a simple answer to what IBM does[1:17]

Arvind's concise description of IBM

Arvind says IBM's role is to help clients improve their business by deploying technology, focusing on business improvement rather than just delivering commodities[1:23]
He emphasizes IBM is not gated to one product; it uses what makes sense at the time[1:29]
At the next layer, he describes IBM as doing this through a mix of hybrid cloud, artificial intelligence, and "a taste of quantum" coming down the road[1:44]
Arvind clarifies he is product agnostic but not technology agnostic[1:53]
He notes that in 25 years IBM could be doing things unrecognizable to today's IBM, such as only open-source software and only quantum computing[2:12]
Malcolm suggests IBM as solving problems at the highest technical level, and Arvind agrees if "highest technical level" is the framing[3:33]

IBM's history of problem-solving illustrated by the barcode

Arvind cites the invention of the barcode by an IBMer as an example of solving a business scaling problem for retailers[2:58]
• He notes the inventor was not a PhD or deep researcher but likely a field engineer
• Lasers were available, but practical issues existed: labels could be upside down, muddy, or scraped, and the IBMer devised the barcode to handle such conditions
Arvind says the barcode changed inventory management forever[4:17]

Arvind's early career at IBM and predicting technological shifts

Starting at Thomas Watson Research Center and building early wireless networking

Arvind joined IBM in 1990 at the Thomas Watson Research Center[4:50]
At that time, computers and networking were beginning to converge, and for his first five years he focused on building networks[5:47]
• He reminds listeners this was pre-laptops; laptops appeared around 1992 or 1993, but portable computing was clearly coming
He spent five years building what we now call Wi‑Fi, iterating designs and making progress[5:47]
• They debated constraints like weight: the Wi‑Fi hardware needed to be under 100 grams so as not to burden a 3,000‑gram laptop
• There was skepticism about why anyone would want to walk around untethered instead of plugging in a thick cable and sitting down, reflecting the terminal mindset of the time

Arvind's 1990 predictions about technology's trajectory

He says he thought primarily about where technology, not industries, would go[5:50]
He would have predicted that networking and computers would fuse, which seemed a niche researcher view in 1990 but was obvious by the late 1990s as the internet[6:47]
He believed video streaming would become the primary way people consume video, and says he would have voiced that in 1990[5:02]
• He notes the technical feasibility existed, but it took about 20 years for costs and usability to make streaming a mainstream reality
Reflecting back to 1985, he says he already considered the internet "old" during grad school because students had networked Apple machines, email, and file transfer, though without a browser[5:20]
• It took about 10 years for the browser to appear and five more for it to become a business, highlighting how consumerization and cost changes lag initial technical use
He notes that after seeing several such cycles, you recognize that in 10-15 years costs drop and consumerization scales technologies in ways previously hard to imagine[5:59]

Broader predictions on digitization of media and on-demand consumption

Arvind says they were convinced linear or broadcast television would become digitized and that putting packet-based television over cable would become the standard[6:20]
He states that as early as 1987 he fundamentally believed on-demand movies would become the way people consume films[6:41]
He did not work personally on all these areas but saw them as easy-to-predict trends given the technical possibilities[6:51]

From invention to markets: blindness, business models, and Arvind's shift toward business

Misalignment between technology potential and business vision in Wi‑Fi

Arvind contrasts his team's view that wireless networking could reach millions to billions of users with the business side's view that the market was limited to warehouse workers doing inventory[7:31]
• When he suggested home users as a market, business leaders could not imagine distribution and buying patterns beyond existing channels
This experience convinced him he could not just help invent technology; he also had to think about marketing, target customers, routes to market, and ease of use[7:48]
He describes this as a 5-10 year evolution of himself toward operating at the intersection of business and technology[7:14]

Why markets are blind to new uses of technology

Malcolm compares this to early telephone history, where adoption lagged for decades because telephone executives resisted women using phones to gossip, failing to see that social use was the product's essence[8:16]
Arvind agrees the pattern repeats: there is often a gap between invention and social understanding of technology[8:36]
He attributes the gap partly to academic disciplines and data-driven thinking that focus on historical data and existing buying patterns[8:53]
• If decisions are based only on past data, they overlook that massive value creators historically built new markets, not just served existing ones
He cites Steve Jobs as an example celebrated for imagining markets that did not exist yet[9:18]
Arvind argues that creating massive value requires three elements together: technology, business acumen to scale a company, and imagination to make new markets[9:29]

Learning business and finance as a technologist

Arvind says that in 1994 he recognized terms like "stock market" and "balance sheet" but had no intuition for them or their relevance[9:56]
He became curious about why companies get higher valuations and how elements like working capital on balance sheets affect that[10:12]
He describes himself as willing to learn and read, but notes he learns finance best by reading actual balance sheets and talking to financial experts rather than just reading textbooks[10:24]
• He found it easier to ask nearby financial experts to explain concepts; they were generally happy to share if he was curious, and over time they became part of his internal network
He frames this learning as both skill-building and network-building within the company[10:46]

Blending scientist mindset with leadership skills

Asked whether successful business leaders must unlearn what made them good scientists, Arvind says he believes the opposite[10:55]
He advocates using one's core strengths as a foundation while deliberately adding complementary skills instead of replacing them[10:53]
He highlights the need for a holistic view: going deep in some areas (e.g., parts of electrical engineering) while maintaining curiosity and moderate depth in areas like finance and marketing[11:27]
He notes leaders must also learn to trust their intuition[11:48]

What Arvind got wrong and lessons from telecom and networking models

Mis-predictions about telecom carriers winning networking

Arvind acknowledges he got many things wrong, including the widespread belief that communication companies (telecom carriers) would be the winners in networking[12:11]
He recalls heavy 1990s investments by telecom carriers, which ultimately did not make them the dominant networking winners[12:14]
He attributes part of their failure to a business model mindset of charging by the minute, inherited from a century of telephony, which clashed with the flat-rate pricing that ultimately won[12:27]

The "stupid network" concept and smart endpoints

Arvind references an article from someone inside a telecom company titled "The Rise of the Stupid Network"[13:05]
He explains that telephone designers believed in a smart network with dumb endpoints (simple phones) that relied on network intelligence for routing and signal processing[13:10]
In contrast, the modern internet is "dumb" in the middle-simply shuttling bits-while the intelligence lives at the endpoints (computers)[13:11]
He suggests this architectural inversion, plus flawed business models, undermined telecom incumbents[13:01]
Asked if he himself thought networks should be dumb or smart in 1990, he says he likely did not think deeply about it but all his work assumed dumb networks that just moved bits[13:39]
• He reasoned that while a smart network might suit voice-only systems, it could not anticipate all future applications, making endpoint intelligence more sensible

Red Hat acquisition: strategy, controversy, and persistence

Perceived heresy: proprietary IBM buying open-source Red Hat

Arvind recounts that in 2018 he proposed IBM acquire Red Hat, an open-source company, even though IBM was known for proprietary software[14:20]
He notes IBM's stock fell 15% on the announcement day, and the deal was widely misunderstood[14:30]
He says today most people consider it IBM's most successful acquisition ever and probably the most successful software acquisition in history[14:40]
He argues people initially failed to see the need for a platform that could be agnostic across multiple clouds and on-premise environments, which Red Hat provides[14:47]

Strategic logic: partnering with cloud leaders instead of chasing them

Arvind describes 2010s cloud dynamics: IBM could have spent large sums (e.g., $10 billion per year) to chase leaders who were already 5+ years ahead[15:20]
He judged that even with heavy investment IBM would likely be a distant third or worse, especially with Chinese cloud providers in the mix[15:41]
Instead of competing directly, he asked whether IBM could occupy a different space by becoming the best partner to cloud leaders and riding their success[15:54]
From that perspective, he sought technologies useful for being a neutral, hybrid platform partner, leading to Red Hat[16:01]

Timeline to conviction and market vindication

He says it took six to nine months of trying and failing to convince people internally of the acquisition logic, followed by about six months of building momentum once a few key people saw it[16:12]
The deal was announced in 2018, took a year to close (with a key date of July 9, 2019 when approvals came), and by around 2023 he feels the world finally granted IBM credit for the move[17:43]
He acknowledges it was a real gamble; had it not worked, he likely would not hold his current role[18:10]

Arvind's temperament: persistence, patience, and sleep

Arvind describes himself as very persistent, very patient, and also somewhat impatient, but says he is not a yeller or screamer[18:29]
He admits his family would certainly describe him as very stubborn and might dispute his claim that he never rants or raves[18:50]
On sleep during the Red Hat bet, he says once the decision was made he did not lose sleep over it[18:59]
• He does wake up about once a week around 2-3 a.m. with his brain running; when that happens he gets up, works productively, accepts being tired later, and sleeps well the following night
He has learned he cannot push intensively across early morning, day, and late night continuously[19:11]
About an hour before bedtime, he deliberately switches to reading something interesting but outside work (e.g., biographies, demographics and population), avoiding leadership or deep science content that would trigger work-related thinking[19:22]
• He looks for material dense enough to fully occupy his brain but not tied to his day job, using it as a gear-shift for his mind

Decision-making style, internal networks, and seeking dissent

How Arvind pressure-tests big ideas like Red Hat

He says he is remarkably open internally and discusses big ideas, including risks as well as benefits, with roughly half a dozen to a dozen people inside IBM[20:22]
He cites specific long-term confidants: current CHRO Nicole (engaged in this loop since at least 2015) and CFO Jim Cavanaugh (since around 2013)[20:48]
• He originally approached Jim to learn finance, asking him to explain concepts he did not understand, which built both knowledge and trust over time
He says many people in the software business have been having such discussions with him "always"[21:17]
While he can get impatient, he insists these are meant to be real discussions where strong differing views are welcome and help refine his own perspective[21:23]

Building a broad community of advisors inside and outside

Arvind suggests each person should build a community of about 100 people inside their enterprise and 100 outside whom they can call for perspective[21:52]
He says he frequently calls external CEOs for five-minute conversations to test ideas, including the former Red Hat CEO who left IBM in 2021[22:07]
• His relationship with that former CEO is mutual; they talk every two or three months and ask each other for opinions, even if the other does something different after hearing the advice
He tells Malcolm that someone like Malcolm, who influences opinions, can provide valuable thinking about how and when technologies like quantum become attractive to a broad audience, even without deep physics expertise[22:50]

Hype cycles, capital allocation, and AI versus the internet

Comparing today's AI moment to the 1990s internet boom

Asked what we are over- or underestimating now, Arvind compares the current moment to the internet around 1995, calling that a major moment bigger than mobile or streaming[23:41]
He recalls that in 1999-2000 people claimed there was excessive hype, but with hindsight the internet more than fulfilled expectations[23:53]
He notes that roughly eight out of ten heavily funded companies went bankrupt, but considers this a positive feature of the U.S. capital system[24:05]
• He emphasizes that assets of failed firms did not vanish; they were acquired cheaply and reused, while the winners (e.g., Amazon and Google/Alphabet) likely repaid the era's total capital many times over
He expects a similar pattern in the current AI wave: many tears but large aggregate success[24:47]
He contrasts the U.S. model with other countries that try to keep all firms alive, which dilutes outcomes and hampers innovation[24:52]
He argues that overwhelming capital deployment when big markets are sensed increases competition, accelerates innovation, and compresses 20‑year progress into five[25:32]

Underused high-ROI AI applications in enterprises

Arvind does not think AI is underestimated globally, given capital and attention, but he believes enterprises often deploy it in the wrong places, chasing shiny experiments instead of scalable basics[26:23]
He advises companies to pick areas they can scale rather than "shiny little toys"[26:39]
He gives two concrete examples of prioritized areas: customer service and software development productivity[26:47]
• On customer service, he says if a company still has more than 10% of the customer service headcount it had 10 years ago, it is already five years behind in using AI
• On software development, he says if a company is not using AI to make developers at least 30% more productive today, aiming for 70% gains, it is falling behind, though he clarifies this means more output, not fewer developers
He estimates only about 5% of enterprises are currently on track in both those metrics[26:23]

AI for the developing versus developed world and demographic pressures

Kenya deforestation use case as an AI blueprint

Malcolm references a prior episode about Kenya's massive deforestation problem where IBM used NASA satellite data and an LLM to produce a 10‑meter-by‑10‑meter map guiding exactly what trees to plant and where[28:08]
He calls it an astonishing ecological blueprint and asks if analysts focus too much on the developed world when AI's greatest ROI might be in developing countries[28:40]

Arvind on AI's dual role across developing and developed regions

Arvind notes software technologies scale as an "and" rather than an "or," serving both developed and developing regions[28:47]
For developing countries, he cites opportunities to greatly improve effectiveness in areas like pesticide and fertilizer use, irrigation targeting, and remote healthcare via AI agents[29:14]
• He gives examples such as moving from flooding fields to precisely irrigating only the plants that need water to achieve 10x effectiveness
In the developed world, he argues societies are running out of people; many regions have birth rates far below replacement[29:20]
• He mentions the Far East may have half its current population by 2070, Europe has sub-replacement birth rates, and U.S. numbers range from 1.6 to 2.1 births per woman depending on how immigration is counted
He concludes that in developed countries AI will be essential to maintain current quality of life because more work must be done despite fewer people[30:06]
In developing countries, AI offers access to advanced capabilities at price points they could never afford otherwise, as Malcolm observes in the Kenyan reforestation project[30:22]

IBM's AI strategy and differentiation from consumer LLMs

Enterprise focus versus consumer chatbots

Arvind stresses IBM is not a consumer company and has no focus on building a B2C chatbot[30:58]
He explains that for B2C chatbots it makes sense to keep making models larger and more computationally inefficient because each new capability can attract millions of users[31:07]
• Examples he gives: adding Finnish or French support, writing haikus, or composing emails in the voice of Steinbeck, each potentially adding around five million users
For enterprises whose goal is tasks like summarizing legal documents in English, he says models can be 1/100 the size, more effective and higher quality without broad capabilities[31:34]
He argues focusing on enterprise tasks reduces the need for extremely large, power-hungry models and massive datasets, easing concerns about copyright and enabling weekend-scale fine-tuning[31:53]

IBM's domain focus and eating its own cooking

IBM focuses its AI on domains where it has deep expertise: operations, programming/coding, customer service and experience, logistics, and procurement[32:38]
Arvind emphasizes IBM applies AI internally first so clients are not the first experiments; IBM shares its own learnings, including necessary process changes[32:53]
He says the biggest AI challenge is not technology but getting people to accept different ways of doing things[33:04]

Challenges of explaining IBM's AI approach and consumer FOMO

He acknowledges it is hard to differentiate IBM's enterprise AI to customers attracted by shiny consumer tools (e.g., people ask why IBM does not have its own GPT-like chatbot)[32:28]
Malcolm mentions a bad experience with ChatGPT where it fabricated about ten people when answering a simple question[33:28]
Arvind generalizes this to all large LLMs, arguing the core issue stems from their reward functions: models learn to produce answers that satisfy users, even by making things up[33:35]
• He likens this to a clever college student who does not know an answer but "bullshits" convincingly, and references the Clever Hans story where a horse appeared to solve problems by reading cues to please its master

Bottlenecks and future directions in AI

Limits of LLMs and the need for knowledge representation

Asked about key bottlenecks, Arvind says he is not convinced LLMs alone will lead to superintelligence or AGI; at best they will see incremental improvements[34:36]
He argues we must find ways to represent knowledge explicitly and fuse it with LLMs instead of statistically rediscovering knowledge from scratch with each query[34:58]

Potential for massive efficiency gains and why they are delayed

On the efficiency front, he believes we can achieve roughly 1000x improvements in power, cost, and compute efficiency for LLMs from today's levels[35:21]
He expects such leaps will drive much greater usage because making something 1000x cheaper massively increases demand[34:36]
He says the path to these gains lies in advances across three fronts: semiconductors, software, and algorithmic techniques[35:36]
He laments that currently, people mostly just scale existing semiconductors and algorithms instead of inventing new ones focused on efficiency[35:21]
He predicts these efficiency-focused advances will occur in less than five years but notes they are delayed because major players prioritizing winning over efficiency are pouring money into brute-force scaling[35:52]
He attributes this dynamic to FOMO: firms fear being left behind if they pause for efficiency while competitors spend heavily to win market share, echoing his earlier internet analogy[36:14]

Quantum computing: a new kind of math and near-term applications

Why quantum is Arvind's favorite topic and how it differs from prior compute

Arvind says quantum computing is his favorite topic and contextualizes it within the history of computation[36:50]
He states that from roughly 1945 (ENIAC) to about 2020 we effectively had one kind of computation: classical[37:02]
GPUs and AI introduced a second paradigm based on neurons and linear algebra, enabling problems hard to tackle on traditional architectures[36:07]
Quantum adds a third kind of math, rooted in abstract algebra and quantum physics, which he says can be thought of via Hamiltonians or Lie algebra[37:28]
He emphasizes that quantum is compute-intensive, not data-intensive like AI; it tackles problems requiring huge amounts of computation[37:39]

Timeline and first signs of "shocking" results

Arvind estimates quantum computing is three to five years away from "shocking" people by doing something they did not think possible in that timeframe[37:34]
He cites a recent example from IBM client HSBC, which published results showing that a quantum computer produced bond trading results 34% more accurate than their prior technique[37:39]
• He notes that in finance, where improvements are measured in single basis points (0.01%), a 34% accuracy improvement is astonishing, even though the experiment was not yet at production scale

Why quantum excels at chemistry and materials problems

Explaining a naive question about batteries, he says equations of quantum mechanics and chemistry are known but lack closed-form solutions, forcing classical methods into massive state-space exploration[38:42]
He illustrates with hundreds of electrons requiring 2^100 states, an impossible memory requirement for classical computing, making such simulations take extremely long[38:28]
Quantum computers, by operating directly in the equation domain, avoid exploring the entire state space and can solve such problems much faster[39:17]
• He suggests quantum could reduce years of computation to a few seconds in predicting material behavior, enabling, for example, design of solid-state batteries with reduced fire risk

Quantum for optimization: routing and fuel savings

Arvind gives a speculative optimization example: a mid-sized country's postal service delivering daily to every address, burning about a billion gallons of fuel per year[40:04]
He notes this can be framed as a traveling salesman problem, which is very hard to solve exactly, leading current systems to use heuristics that may be 20% off optimal[40:45]
If quantum could improve routes by even 10%, that would save roughly 100 million gallons of fuel (from a billion), translating in his example to about 800 million pounds in savings for one entity in one year[40:04]
• He points out additional uncounted benefits: reduced carbon footprint, fewer emissions, less wear and tear, and lower vehicle mileage
He mentions interest from regions like New York and Illinois, which have launched quantum algorithm programs and centers involving IBM and universities[40:50]
He says IBM has roughly 200 clients doing early-stage experiments because they intuitively see quantum can solve problems classical methods cannot[40:50]

Quantum's place among great inventions

Asked to rank quantum among major inventions of the last 150 years, Arvind places it equal to the semiconductor[41:28]
He observes that if semiconductors vanished, modern life-electricity-dependent systems, automobiles, streaming-would stop[41:20]
He argues quantum is under-discussed because it is still four to five years from its "browser moment"-an accessible abstraction akin to Netscape for the internet[41:47]
He notes that talking about new kinds of math or quantum mechanics immediately loses most of the audience, further slowing mainstream recognition[42:00]

Arvind's role in IBM's quantum journey and leadership approach

Origin and scaling of IBM's quantum efforts

Arvind says he began investing in quantum around 2015 when leading IBM Research, after a career that moved between research, acquisitions, and business strategy[42:31]
Initial goals were modest: make a computer, not just a science experiment; get it to run autonomously overnight; and build software so non-quantum specialists could use it[42:50]
Over three to four years, accumulating successes in these areas gave him confidence that quantum could really work and warranted growing investment[43:41]
He says the moment he regarded quantum as a big bet probably came two or three years before this conversation[43:41]

Allocating resources and managing pressure on frontier projects

Arvind describes three considerations for resource allocation: availability of truly expert people, risks of pushing too hard on timelines, and the art of setting achievable yet ambitious expectations[43:43]
He notes he is constrained by the limited number of people with deep quantum expertise and would hire more if possible[43:43]
He cautions that if leadership pushes too hard on timing, teams might take excessive risks, commit impossible timelines, and cause projects to fail[43:58]
To hit a Goldilocks level of pressure, he emphasizes openness: teams must feel free to push back hard, argue, and adjust goals realistically[45:03]
He also wants project leaders whose personalities drive them to go as hard as possible without crossing into the impossible[44:41]
He believes the quantum leadership team is comfortable arguing with him, so he has not felt he has pushed too hard at any point[45:03]
He checks on competitive awareness and initiative by texting people questions about competitor moves or scientific developments and seeing if they can thoughtfully answer, which reassures him they are already on top of things[45:15]

Closing reflections on Arvind's job and purpose

Is this the most interesting job in America?

Asked if he has the most interesting job in America, Arvind says he believes it is the most interesting job and that he would not give it up for anything[46:19]
He says he enjoys the role as long as he can make the enterprise thrive and clients delighted; if he cannot, someone else should do it[45:22]

Farewell and looking ahead on quantum

Malcolm thanks Arvind for the conversation, calling it fascinating and joking that he wishes he could help with quantum but cannot yet[46:53]
Arvind replies, "In a few years," implying quantum will become more accessible to a broader set of people over time[46:53]

Lessons Learned

Actionable insights and wisdom you can apply to your business, career, and personal life.

1

Technological innovation creates the most value when it is paired with a deep understanding of business models, customer behavior, and routes to market, rather than focusing on invention in isolation.

Reflection Questions:

  • • Where in your work are you focused on building or improving something without equally considering who will use it, how they'll access it, and why they would pay for it?
  • • How could you bring in people with strong business or customer insight to stress-test the commercial potential of your technical or creative ideas?
  • • What is one current project where you could explicitly map out the target customer, distribution channel, and pricing model this week?
2

Building a broad, trusted network of internal and external advisors dramatically improves decision quality, especially for high-stakes, unconventional bets.

Reflection Questions:

  • • Who are the 5-10 people inside and outside your organization that you can call today for an honest, informed perspective on your toughest decisions?
  • • How might your recent big decisions have been different if you had proactively sought out dissenting or complementary viewpoints?
  • • What concrete step could you take in the next month to start assembling your own "community of 100" advisors over time?
3

Persistent, patient leadership means pushing hard on conviction-driven bets while creating psychological safety for teams to push back, adjust timelines, and surface risks.

Reflection Questions:

  • • On which current initiative are you so convinced you might unintentionally be discouraging honest pushback from your team?
  • • How could you explicitly invite disagreement and risk reports in a way that makes people feel safer rather than threatened?
  • • What is one project where you may need to recalibrate pressure-either by raising expectations or easing timelines-to reach a realistic yet ambitious "Goldilocks" zone?
4

To stay relevant as a technical expert, you must continuously extend your skill set into adjacent domains (like finance, marketing, or operations) instead of relying solely on your original specialty.

Reflection Questions:

  • • Which non-technical domain-such as finance, marketing, or operations-do you currently understand least but which strongly influences your success?
  • • How could you learn more effectively by studying real artifacts (e.g., balance sheets, customer journeys) and asking domain experts questions instead of only reading abstract material?
  • • What specific cross-disciplinary skill or concept will you commit to learning over the next 90 days, and who can help you accelerate that learning?
5

Organizations gain more from AI when they target high-impact, scalable use cases-like customer service and developer productivity-rather than chasing flashy experiments with limited business value.

Reflection Questions:

  • • Which repetitive, large-scale processes in your organization (support, coding, logistics, documentation) might benefit most from AI augmentation right now?
  • • How can you reframe your AI initiatives from "cool demos" to measurable productivity or cost improvements that matter to your core business?
  • • What is one specific AI pilot you could scope this quarter that has a clear path to scaling across your organization if it works?
6

Preparing early for paradigm shifts like quantum computing or major AI shifts creates a future advantage, even when the technology is not yet fully ready for production.

Reflection Questions:

  • • What emerging technologies are likely to reshape your industry in 3-10 years, and how familiar are you with their basic principles today?
  • • How might small exploratory projects or partnerships now reduce your risk of being disrupted when those technologies mature?
  • • What is one low-risk way you could start building hands-on familiarity with a transformative technology-such as a pilot, training, or collaboration-over the next six months?
7

In aging, labor-constrained societies, intelligent automation is not just an efficiency play but a necessity to maintain quality of life as the available workforce shrinks.

Reflection Questions:

  • • Where is your team or organization already feeling strain because there simply are not enough people to do all the needed work?
  • • How could you reimagine roles, processes, or tools so that machines handle more routine tasks and people focus on higher-value work?
  • • What is one area of your operations where you could deliberately pilot automation or AI to alleviate workload without compromising quality?

Episode Summary - Notes by Avery

IBM CEO Arvind Krishna: Creating Smarter Business with AI and Quantum
0:00 0:00