The Spark That Ignited The Current AI Revolution -- May 6, 2026

Query: what was the spark that occurred 2016 - 2020 that was ignited the AI revolution in 2022 (post-Covid)?

Reply:

 

Industrial Revolutions -- May 6, 2026

Query: Industrial revolutions. The first industrial revolution was steam. The second industrial revolution was railway and much more. How long did the second industrial revolution last?

 

 

Query: 

I would argue the current industrial revolution began after Covid -- Nvidia broke out from gaming to AI in 2001. When would you same the current AI revolution began? The second industrial revolution lasted 44 years; how long might the AI revolution last? 

Reply:

 

LNG Export Chokepoints -- May 1, 2026

 Query: 

The US has too much natural gas; the world not enough with the Iran War -- what's the major chokepoint getting US LNG to Europe, Asia? US pipeline grid (continental); the last mile LNG pipelines in Texas, Louisiana; the export terminals along the Gulf Coast; tankers; existing long-term contracts; something else?

Reply:  

War Powers Act

The War Powers Act is not specifically "enacted" by anyone. It simply takes effect by statute. The courts do not start the process, but the executive branch can be sued by a member of Congress if the member feels the administration is not abiding by the law. Historically, the courts do not get involved. The US Supreme Court considers this a political issue and they won't get involved; this is pretty much an ironclad issue for SCOTUS.

US Minesweepers -- April 25, 2026

 

Lucent In 2000 -- Is History Repeating Itself? April 25, 2026

AI query: Is the euphoria over AMD, NVDA, MU similar to the euphoria over Lucent in 2000?

 MIT Technology Review: how Lucent lost it. Link here.

 Tom Tunguz: Nvidia is not Lucent. Link here.

*************************
Lucent Imploded 2000 - 2001

AI query: what were the key highlights of Lucent's financial report for the year 1999: market cap, free cash flow, debt, eps, revenue, revenue growth year-over- year. 

Lucent, 1999:

 

AI query: 2hat were the key highlights of Cisco's financial report for the year 1999: market cap, free cash flow, debt, eps, revenue, revenue growth year-over- year.

 

Cisco, 1999: 

Richest Eight Families In Forth Worth, Texas -- April 14, 2026

l

Link here. 

Apple, Amazon Leo, And Globalstar — April 14, 2026

Link here. 

Apple has a deep, multibillion-dollar partnership with Globalstar, initiated in 2022 to power the Emergency SOS via Satellite feature on iPhones. Through significant investments and prepayments exceeding $1.5 billion, Apple secured 85% of Globalstar's network capacity, utilizing it for satellite messaging and emergency services.

Key Details of the Partnership: 

• Infrastructure Investment: Apple's funding allows Globalstar to expand its ground infrastructure and upgrade its satellite constellation. 
• Capacity Control: Under the agreement, 85% of Globalstar’s network capacity is reserved for Apple services. 
• Financial Support: Apple has provided over $1.5 billion in payments, including funding for new satellites and supporting Globalstar’s operational costs. 
• Service Scope: The partnership enables iPhone 14 and newer models to connect directly to satellites to send emergency texts and location data in areas without cellular coverage. 
• Recent Developments: In November 2024, Globalstar announced a $1.7 billion investment from Apple to launch a new, enhanced constellation. 
• Ownership Stake: Apple holds a 20% stake in Globalstar, strengthening the long-term partnership.  

Amazon-Globalstar Merger (April 2026): As of April 2026, Amazon is set to acquire Globalstar for $11.6 billion. Despite this merger, reports indicate the Apple-Globalstar agreement remains intact, and Apple will continue to use the network for its satellite services. 

Other Collaborations: While Globalstar is the primary partner, Apple has also explored, or currently works with, other providers for satellite technology to expand beyond emergency-only services, though Globalstar remains central to their current strategy. 

Amazon Leo:

Amazon Leo is Amazon’s low Earth orbit (LEO) satellite broadband network, formerly known as Project Kuiper, designed to provide high-speed, low-latency internet to households, businesses, and governments worldwide. Utilizing over 3,000 satellites, it aims to connect underserved areas, with service options planned for in-flight Wi-Fi and direct-to-device connectivity.

Micron -- April 13, 2026

Links: four important links below. 

Micron could surge another 40%. Link here. 
The queue for storage. Link here.

• Nvidia
• Apple
• AWS

Largest fab being built in the US. Link here.

Revenue: google -- micron revenue quarterly history 5-year

******************************
Micron vs Sandisk

AI prompt: I think it can be safe to say that Micron and Sandisk are not in direct competition. Though both are in the storage business, they are in different markets. Is that correct?

Reply: 

How AI Changed Investing In One Chart -- Axios: Sandisk Replaces Atlassian On The NASDAQ 100 -- Monday, April 13, 2026

How AI changed tech stocks, in 1 chart

Data: Financial Modeling Prep; Chart: Emily Peck/Axios Memory chipmaker Sandisk is replacing workplace software maker Atlassian on the Nasdaq 100 index, Nasdaq announced late Friday. Why it matters: The small move is a big symbol of what has happened this year as the tech industry adapts to the AI boom. Zoom in: Software companies are getting crushed by the competition in the SaaS-pocalypse, while hardware makers are thriving. Sandisk makes flash memory storage chips that are essential to data centers. The company spun off last year from Western Digital and has been on a tear ever since — rising to $851.77 from $36 in February 2025, when it made its market debut. The bottom line: AI has made software uncool and hardware the belle of the ball.

ASU -- Arizona State University Center -- Will Develop Its Eighth Arizona "Innovation Center" -- To Be Near TSMC's $165 Billion Plant -- April 5, 2026

From The Arizona Republic, April 5, 2026.

Ari­zona State Uni­versity will develop its eighth Ari­zona “innov­a­tion zone,” an area that will be devoted to inter­na­tional sci­ence and col­lab­or­a­tion, near the Taiwan Semi­con­ductor Man­u­fac­tur­ing Co. facil­ity in north Phoenix.

The current seven Arizona "innovation zones": 

ASU Pres­id­ent Michael Crow called the $165 bil­lion TSMC plant near Inter­state 17 and Loop 303 the “single most import­ant tech­no­lo­gical oppor­tun­ity in Ari­zona since air con­di­tion­ing.”

“We are going to con­cen­trate an innov­a­tion dis­trict there that’s linked to this inter­na­tional sci­ence and tech­no­logy com­munity,” Crow said.

The uni­versity appoin­ted Grace O’Sul­li­van as vice pres­id­ent of TSMC.

Part­ner­ship Ini­ti­at­ives, and she works closely with TSMC and semi­con­ductor industry sup­pli­ers to address their needs for work­force, research and uni­versity part­ner­ships.

O’Sul­li­van said she is work­ing with Crow and other ASU lead­ers to design the innov­a­tion zone, called the Inter­na­tional Sci­ence Zone.

“We really see Ari­zona as becom­ing that global land­ing spot,” she said about the area’s abil­ity to attract inter­na­tional com­pan­ies.

The innov­a­tion zone will give ASU an oppor­tun­ity to cre­ate an advanced teach­ing and learn­ing cen­ter and joint aca­demic cen­ters with inter­na­tional uni­versit­ies, O’Sul­li­van said.

The zone will also have space for star­tup incub­a­tion, which O’Sul­li­van said is a gap in the mar­ket for advanced man­u­fac­tur­ing and tech­no­logy com­pan­ies. The zone could also include other edu­ca­tional entit­ies like com­munity col­leges or K-12 schools, along with other facil­it­ies the com­munity desires, like recre­ation.

“We want to cre­ate a hol­istic work­force devel­op­ment sys­tem,” she said.

The uni­versity will need “quite a bit of ter­rit­ory” for the zone and that it is “look­ing every­where” for land but Crow did not con­firm a spe­cific loc­a­tion.

Crow said the uni­versity is mov­ing at “flank speed” on get­ting the innov­a­tion zone going, a naut­ical phrase mean­ing as fast as the boat can go.

ASU has seven other “innov­a­tion zones,” which are des­ig­nated areas where the uni­versity has oper­a­tions along­side private busi­nesses. Those loc­a­tions include three nearby the main cam­pus in Tempe: The Novus Innov­a­tion Cor­ridor, ASU Research Park and SkySong in south Scott­s­dale.

The other Ari­zona innov­a­tion zones are the Dis­cov­ery Oasis Health Futures Cen­ter near the Mayo Clinic in north Phoenix, the Phoenix Bios­cience Core in down­town Phoenix, the ASU Mesa Cen­ter for Cre­at­ive Tech­no­logy in down­town Mesa, the ASU Poly­tech­nic Innov­a­tion Zone in Mesa and the ASU West Val­ley Innov­a­tion Zone in west Phoenix.

Cor­ina Vanek cov­ers devel­op­ment for The Ari­zona Repub­lic. Reach her at cvanek@ari­zon­are­pub­lic.com. Fol­low her on X @Cor­inaVanek.

Evanston, Lordstown, SoftBank -- A Geography Lessson -- February 20, 2026

How close is the proposed Lordstown LDC and proposed SoftBank power station to Evanston, IL?

Stargate, Ohio, Softbank, LDCs, Powerplants -- February 19, 2026

Background:

Not only the largest natural gas power plant but the largest power plant, of any sort -- nuclear, coal, natural gas -- in the entire United States. 

Wiki: list of largest US power plants. 

• Grand Coulee, hydro, 6,809 MW
• Alvin W Vogtle, nuclear, 4,536 MW, 
• Palo Verde, nuclear, 3,942 MW,
• Browns Ferry, nuclear, 3,775 MW,
• West County (Florida), natural gas, 3,750 MW,
• W. A. Parish (Texas), natural gas / coal, 3,653 MW

SoftBank proposal:

 

The Brilliance Of Trump -- February 19, 2026

Harvard is in severe financial distress.

• Trump's brilliance: how he "read" Harvard's Achille's heel;  
• the union killing the Golden Goose; no sympathy

 

Vannevar Bush -- February 16, 2026

AI prompt:

Alan Turing. Development of computers. Things don't happen in a vacuum (no pun intended). Things don't happen overnight. Specifically, 1937 was a pivotal year and not just with Turing. From 1843 (Ada) to 1931, Vannevar Bush, almost a 100 years -- not much happening in those 100 years. Then Tommy Flowers in 1935 -- valves / vacuum tubes.  So you look back five to six years before that pivotal year. What was happening -- or better, perhaps -- what major geo-political-economic event happened five to six years earlier. In fact, it was about eight years but one gets the point: the market crash of 1929. I'm sure there are some observations / thoughts about those years ... 1929 to 1937 that brought us to that point. Thoughts?

Google Gemini: 

Norway's Wealth Tax -- February 13, 2026

Even without the wealth tax, I would prefer to live in Switzerland rather than in Norway.

The big problem is that the 1.1% wealth tax on net assets over $US 2 million is just part of the overall tax these two-million-dollar (US dollars) millionaires are paying. 

Even worse, $US 1 million is no longer all that much wealth in the big scheme of things.  

The big question is whether Norway has either a debt problem or a deficit problem:

Norway's wealth tax: 

Norwegian wealth tax. Story reported in The Guardian. AI prompt and reply:

 

In fact, whether it was good or bad / increase or decrease in tax revenue in short term, it is clearly a huge negative in the long term, for many reasons:

*************************
Wealth Tax

If the wealth tax is a detriment to the economy in a country like Norway, one can only imagine how bad the wealth tax would be to a state like California who is at risk of a major deficit in any given year.

Fourth Industrial Revolution -- Google Gemini -- February 11, 2026

 

Fourth Industrial Revolution -- ChatGPT -- February 11, 2026

My thesis: for the mom-and-pop retail investor, the tipping point for investing "within the fourth industrial revolution" was in 2021 - 2022 -- a year or so after then "end" of the Covid-19 anxiety even though technically the fourth industrial revolution was identified in 2015 - 2016.

ChatGPT: 

Yeah—that intuition tracks. The conceptual “Fourth Industrial Revolution” (AI, automation, digital platforms) went mainstream around 2015–2016, but the retail-investor tipping point didn’t really hit until 2021–2022. Different clocks were ticking for institutions vs. everyday investors. A few reasons why your timing feels right:

⸻

Why 2016 ≠ the retail tipping point

2016 was narrative-level adoption.
That’s when AI became a serious boardroom topic (DeepMind/AlphaGo, early cloud AI services, etc.). But for mom-and-pop investors, the story was still abstract: “Cool tech, not sure how to play it.” Public-market exposure was limited to broad tech names and semis.

Retail tends to move when:
    •    products become visible in daily life, and
    •    returns become obvious in stock charts.

Neither was true yet in 2016.

⸻

Why 2021–2022 was the tipping point

1. COVID changed investor behavior, not just tech adoption

Lockdowns did three huge things at once:
    •    Time: people had more time to learn markets
    •    Access: Robinhood/zero-commission trading went mainstream
    •    Psychology: markets felt “democratized” (GameStop, AMC, crypto)

This created a brand-new retail cohort that:
    •    was younger,
    •    more speculative,
    •    and much more willing to chase narratives (AI, EVs, space, quantum, etc.).

AI benefited from that same narrative trading energy that fueled meme stocks.

⸻

2. The tech finally crossed from “infrastructure” to “products people could feel”

Before 2021, AI mostly lived behind the scenes:
    •    ad targeting
    •    search ranking
    •    fraud detection
    •    logistics optimization

After COVID:
    •    recommendation engines, automation, and bots were suddenly everywhere
    •    then ChatGPT (late 2022) made AI personally tangible

That’s when retail investors went from:

“AI is a thing companies use”
to
“Oh wow, this changes everything.”

That emotional shift matters a lot in retail flows.

⸻

3. NVIDIA as the retail “AI thermometer”

Retail investors don’t invest in systems—they invest in symbols.

NVIDIA became the symbol:
    •    clean story
    •    obvious revenue growth
    •    simple narrative: “AI runs on GPUs”

Once NVDA started ripping in 2023, it functioned like:

“Proof that AI isn’t just hype—this is how you get paid.”

That feedback loop (price → attention → more retail inflows → price) is classic retail tipping-point behavior.

⸻

4. Post-COVID narrative hunger

After:
    •    meme stocks cooled,
    •    crypto crashed,
    •    EV hype got messy,

retail investors were hungry for a new mega-theme that felt:
    •    transformative
    •    understandable
    •    not already “over” (psychologically)

AI checked all three boxes.

⸻

Why institutions moved earlier than retail (and always will)

Big money started positioning in:
    •    cloud,
    •    data centers,
    •    semis,
    •    ML tooling

years earlier—quietly. Retail tends to arrive when:
    •    revenue curves bend upward, and
    •    products enter culture.

That’s not stupidity—it’s just different incentives:
    •    institutions invest in capability
    •    retail invests in visibility



All four together created what feels like an overnight AI boom—but it was actually a slow fuse with a loud detonation.

⸻

Big-picture take (your instinct is strong here)

What you’re really noticing isn’t just “AI became a thing” — it’s:

When the story became tradable for regular people.

That’s the true retail tipping point.
Not when the tech exists.
Not when it’s useful.
But when it becomes:
    •    narratively simple,
    •    culturally visible, and
    •    attached to stocks that go straight up.

If you want, next time we can map this exact same pattern onto:
    •    the internet (late 90s),
    •    smartphones (2010–2012),
    •    cloud computing (2014–2016),
    •    crypto (2017 / 2021).

The cycles rhyme a lot.

Palantir -- February 2, 2026

AI prompt:

With the announcement today that SpaceX (and xAI will merge, suddenly, it's gotten very, very interesting. Think about this: the Pentagon (DOD), all branches of the US military, NSA, NASA, SpaceX (Spacelink / Starlink) .... the one software company that connects them all? Palantir. 

ChatGPT:

 

🔥 What Happened: SpaceX 

and

 xAI Are Now One Company

Today Elon Musk announced that SpaceX has merged with his AI company xAI, forming a combined entity with an estimated ~$1.25 trillion valuation — and that this consolidated company is aimed at combining space technology, AI, communications platforms, and data infrastructure into a single vertically integrated powerhouse. 

Key points from the merger:

• SpaceX acquires xAI and subsumes its AI assets (including the Grok chatbot and social platform X) under the SpaceX umbrella. 

• A strategically stated goal is to build large-scale AI data centers in space — potentially using solar-powered satellites to provide enormous computing capacity without the terrestrial power and cooling constraints of today’s data centers. 

• The combined company is positioning itself to compete with major cloud + AI players (like Google, OpenAI, Meta, and others) in both compute and deployment infrastructure. 

This blurs the lines between space infrastructure, AI compute, and communications in a way few expected.

---

🚀 Why This Makes the Palantir Angle So Interesting

You mentioned something insightful: SpaceX + xAI intersects the realms of government/military infrastructure, satellite communications, and AI — a territory where Palantir is deeply embedded.

Here’s how Palantir plays into this:

🧠 1. Palantir is 

already

 deeply embedded in defense data and AI tooling

Palantir’s software (such as Gotham and Vantage) is used extensively across Pentagon, Army, intelligence agencies, and allied defense systems to consolidate, analyze, and operationalize massive data streams. They help users from logistics to battlefield insights and AI-assisted decision-making across departments. 

💼 2. Defense contracts are a major driver of their growth

Palantir reported significant increases in U.S. government and military contracts, with revenue and forecasted growth supported by deals with the Department of Defense, Army, Navy, and more. 

Palantir’s government business alone has grown substantially — and reporting indicates high-value, multi-billion-dollar defense contracts which tie it directly into U.S. national security infrastructure. 

🛰️ 3. Partnerships between defense tech firms — 

including SpaceX and Palantir

 — have already been reported

There are reports of coalitions of defense tech companies — Palantir, Anduril, SpaceX, OpenAI, and others — bidding for Pentagon contracts together. While those discussions aren’t necessarily formal mergers, they illustrate how these companies are increasingly grouped strategically from a government procurement perspective. 

This suggests Palantir isn’t just a standalone software contractor: its data tooling is part of how the U.S. military integrates and operationalizes AI across sensors, satellites, and ground systems.

---

📡 SpaceX–xAI Merger and Palantir: Different Roles, 

But Potentially Complementary

Let’s map out the rough landscape:

SpaceX/xAI

• Infrastructure provider (rockets, satellites, Starlink)

• AIMING to become a global AI compute backbone

• Platform for distributed communications

• Has Pentagon ties (Space Force, military contracts)

• Positioned for a huge IPO

Palantir

• Data integration and analytics platform

• Deeply used by the Pentagon and intelligence community

• Enables decision-level AI and data workflows across agencies

• Not selling compute hardware, but valuable software layers that sit atop it

So: they aren’t direct competitors.

Rather, they could be different layers of the same machine — if government and commercial defense buyers want end-to-end systems that run from data capture (satellite or sensor networks) → inference and AI (compute) → analysis and decision support (software like Palantir).

That’s a powerful combo: hardware infrastructure + AI compute + operational software.

---

🧭 Why It Matters to National Tech Strategy

This merger isn’t just another corporate play — it speaks to broader strategic trends:

✅ 

Integration of AI with global communications and defense infrastructure

• SpaceX’s Starlink already serves military connectivity

• xAI’s AI pushes computational boundaries

• Palantir’s software stitches disparate data into actionable insights

If you imagine a future battlefield, space surveillance, logistics operations, or intelligence fusion center — all of these capacities are increasingly essential and interconnected.

✅ 

Government and defense budgets are now major drivers of AI adoption

Beyond consumer or enterprise cloud usage, national strategy (AI used in defense, intelligence, security) is a massive and rapidly growing bucket of spending. Palantir is currently one of the principal commercial providers in that space. 

✅ 

New forms of competition and collaboration

Today’s competition isn’t just “GAFA vs. OpenAI” — it’s platform stacks that intertwine physical infrastructure (satellites, networks), compute, models, and operational AI. SpaceX’s move toward space-based data centers and AI computing is an example of pushing the hardware envelope, which could dramatically reshape competitive economics for hyperscale compute.

In that context, a software integrator like Palantir (which turns raw data into decisions) becomes significantly more relevant.

---

🧠 Final Thought

Your instinct that Palantir is uniquely positioned was on point — but it’s not that Palantir currently “connects all those dots” in a single stack by itself. Rather:

• Palantir’s software enables data and AI at scale in defense and intelligence

• SpaceX/xAI’s merger is building the physical and compute infrastructure for next-generation AI

• Together they map onto different layers of how tomorrow’s strategic systems will actually operate

So yes — systemically, we’re talking about AI, defense, space, and data infrastructure all converging. Palantir sits in the software layer that makes sense of data from satellites, sensors, and networks — and today’s SpaceX-xAI deal dramatically reinforces why those layers matter together, even if the companies themselves are separate entities.

Let’s watch how customers like the Pentagon or NASA begin to think about procurement — that’s where these stacks become real.

 

CAT -- January 29, 2026

This is a great case study, because Caterpillar looks like “old industrial America” on the surface, but its energy pivot was neither luck nor sudden. It was slow, board-level, and frankly very unsexy—which is why it worked.

Short answer: CAT began realizing “energy” was a strategic pillar in the late 2000s, made it explicit in the early 2010s, and quietly locked it in during 2015–2020. By the time investors noticed, the heavy lifting was already done.

Let me give you a clean timeline and then answer the luck vs. foresight question.

---

Phase 1: The seed is planted (late 1990s–2008)

This is where most people miss the story.

Caterpillar already had:

• diesel engines (Cat Engines)

• gensets

• marine and industrial power systems

But these were treated as adjacent businesses, not the core identity.

Two things changed their internal thinking:

1. Globalization of energy demand

   • Mining, oil & gas, and infrastructure all scaled together

   • Customers didn’t want “machines” — they wanted power + uptime

2. Margin realization

   • Engines, power systems, and services had:

     • longer lifecycles

     • higher service margins

     • less cyclicality than construction equipment

By the mid-2000s, internally, CAT already knew:

“Power is stickier than iron.”

---

Phase 2: Post-GFC clarity (2009–2013)

The Global Financial Crisis was pivotal.

Construction collapsed.

Mining whipsawed.

But energy demand didn’t disappear — it shifted.

This is when CAT leadership started explicitly talking about:

• “Energy & Transportation” as a reporting segment

• Lifecycle value

• Installed base monetization

Key signal (often overlooked):

• CAT began investing heavily in aftermarket, service contracts, and remote monitoring

• This only makes sense if you believe power assets will run for decades

This was not luck.

It was defensive strategy informed by pain.

---

Phase 3: The Bucyrus deal (2011) — a warning shot, not the pivot

The $8.8B acquisition of Bucyrus (mining equipment) is often cited as a mistake.

But strategically, it taught CAT something critical:

• cyclicality kills capital returns

• customers value reliability over novelty

• energy-adjacent assets survive downturns better

After Bucyrus, CAT:

• became much more disciplined on capital allocation

• leaned harder into engines, turbines, power systems, and services

This is when the board really internalized the lesson.

---

Phase 4: Explicit energy framing (2014–2018)

This is when CAT stopped hinting and started structuring around energy.

Concrete moves:

• Formal Energy & Transportation segment

• Expanded natural gas engines

• Microgrid solutions

• Power for remote/off-grid sites

• Marine and rail electrification hybrids

Crucially:

CAT did not frame this as “green” or “ESG.”

They framed it as:

• reliability

• resilience

• fuel optionality

That let them sell to everyone.

---

Phase 5: Quiet payoff (2019–2023)

By the time:

• data centers needed backup power

• utilities needed fast-deploy generation

• mining electrification accelerated

• LNG infrastructure expanded

CAT was already embedded.

To outside investors, it looked like:

“Wow, Caterpillar got lucky with energy.”

Internally, it looked like:

“We’ve been building this for 15 years.”

---

So… lucky or prescient?

Not luck.

But also not a single “aha” moment.

This was institutional pattern recognition, not founder genius.

CAT’s advantages:

• Long planning horizons

• Engineering-first culture

• Board comfort with capital intensity

• Willingness to be boring while building option value

They didn’t predict the energy transition.

They positioned themselves so that almost any energy future benefits them.

That’s the key.

---

Why this matters 

now

CAT is quietly becoming:

• a beneficiary of data center buildout

• a supplier to grid resilience

• a player in distributed generation

• a services-heavy energy company disguised as an equipment maker

They don’t need AI hype.

They sell power when power is scarce.

In today’s cloud + AI + data center world, that’s not incidental.

---

Final framing

Caterpillar didn’t pivot to energy.

They realized:

“Energy was always our highest-quality business — we just hadn’t admitted it yet.”

That realization likely crystallized between 2009 and 2013, and everything since has been execution.

If you want, next time we can compare this to:

• GE (what went wrong)

• Siemens (what partially worked)

• or why CAT’s culture made this possible when others failed

AI prompt: 

That is interesting. I had not thought of that. What happened to GE (similar to what happened to Intel)? So, two questions: what went wrong for GE and was it similar for Intel or something completely different?

 

This is one of those comparisons that looks obvious on the surface (“big American industrial icons stumble”) but turns out to be really instructive once you peel it back. GE and Intel failed in different ways, but they share a deep structural rhyme. Caterpillar avoided both traps, which is why it’s such a useful contrast.

I’ll take them one at a time, then tie them together.

---

What went wrong at GE

The short version

GE didn’t die because it couldn’t build things.

It died because it forgot what its core competence actually was.

The long version (but still tight)

1. Financialization ate the company

GE Capital went from:

• supporting industrial sales

  to

• becoming the profit engine

By the late 1990s and early 2000s:

• earnings smoothing mattered more than engineering truth

• risk was misunderstood, not measured

• the board got addicted to predictability

This broke the feedback loop between reality and reported performance.

---

2. Complexity outpaced control

GE became:

• aviation

• power

• healthcare

• oil & gas

• finance

• media

• appliances

Each business was defensible.

The system was not.

The CEO role became:

“Allocate capital across black boxes”

That works—until it doesn’t.

---

3. Long-cycle businesses were managed like short-cycle ones

Power turbines, jet engines, and grid infrastructure:

• have decades-long horizons

• require brutal honesty about reliability and maintenance

GE leadership kept applying:

• quarterly targets

• financial engineering

• cost pressure

Reality eventually asserted itself (see: power division collapse).

---

4. Cultural decay

GE’s famous management system turned from:

• talent development

  into

• political survival

Bad news didn’t travel upward.

That’s always fatal in industrial companies.

---

What went wrong at Intel

Intel’s failure looks different—but it rhymes.

The short version

Intel lost process leadership and didn’t know how to live without it.

---

Key failures

1. Over-internalization

Intel believed:

“We are our process.”

That was true—until it wasn’t.

When they fell behind:

• they couldn’t outsource without identity damage

• they couldn’t admit delay without credibility loss

TSMC ate their lunch while Intel argued with itself.

---

2. Culture optimized for incrementalism

Intel was brilliant at:

• optimizing known architectures

• extracting yield improvements

They were bad at:

• disruptive change

• killing sacred cows

• reorganizing around new constraints

This is deadly in semiconductors.

---

3. Capital intensity without flexibility

Fabs require:

• perfect execution

• constant reinvestment

• external discipline

Intel had:

• legacy customers

• internal politics

• government entanglements

That’s a rough mix.

---

4. CEO degrees of freedom collapsed

Like GE, Intel reached a point where:

• the CEO’s job was survival, not vision

• every move upset a constituency

• strategic optionality vanished

By the time Pat Gelsinger arrived, the game was already constrained.

---

The similarity (this is the key insight)

GE and Intel both suffered from identity lock-in.

GE thought it was:

“The world’s best managed company”

Intel thought it was:

“The world’s best chipmaker because of process leadership”

When those identities broke, neither had:

• fast feedback loops

• cultural humility

• structural escape hatches

Caterpillar did.

---

The crucial difference with Caterpillar

CAT never confused:

• what it builds

  with

• why customers need it

GE and Intel both did.

CAT always knew:

“Our job is uptime, reliability, and service—whatever the energy source or machine.”

That let them evolve without existential crisis.

---

So are GE and Intel “the same story”?

No—but they rhyme.

Same ending mechanism:

Reality catches up faster than the organization can adapt.

---

Final thought

The lesson isn’t “don’t get big.”

It’s:

Never let identity outrun feedback.

Once a company can’t hear bad news, can’t pivot without self-harm, or can’t admit the ground shifted—it’s already late.

Caterpillar stayed boring, honest, and close to physics.

GE and Intel drifted into abstractions.

That difference explains almost everything.

If you want, next time we can look at:

• Microsoft’s near-miss (Ballmer era)

• IBM’s slow survival

• or whether today’s AI giants are drifting toward GE-like risk