Foreboding: Technology is at a hinge point.

I still have the willies.

Artificial intelligence companies are "too big to fail." I have heard that before. 

Guest post author John Coster asks if this moment reminds us of the year 2000, the year the internet bubble popped. It does for me.

The frenzied build-up of data centers reminds me of the frenzied build-out of fiber cables. The stories of fantastic potential uses of artificial intelligence remind me of the presumed remaking of the economy by the internet. (It happened, but not on schedule.) But mostly I am reminded that technology platforms and infrastructure need to be paid for with profits, which wasn't happening in 2000 and may not happen now, not on schedule at least. Artificial intelligence platforms need to create income to meet their goal of going public; AI customers are getting surprise charges for using AI and they are realizing that they need to start shopping for utility and price.  Users are learning that Chinese platforms can do the same work as the well-known frontier platforms for a fraction of the cost.

The crisis is one common in technology: It gets cheaper. WAY cheaper, and then it becomes free. 

John Coster looks back 30 years to the prior technology bubble. We learn from history, but but we make the same mistakes again anyway. 

Coster enjoyed a 45-year career planning, building and operating critical facilities for major technology firms, as well as the energy, education, healthcare, and heavy manufacturing sectors for Fortune 50 companies. He currently advises and consults with venture capital clients focused on emerging technologies. He holds multiple patents using AI for energy optimization in high-tech facilities.

Guest Post by John Coster

Does It Feel a Little Like 2000 All Over Again? 

Back in the late 1990s, when I was running Microsoft's small (by today's standards) portfolio of data centers, massive amounts of IPO money were being poured into what SoftBank's founder called the "Internet zaibatsu." It was a thrilling vision: a global, interconnected web of technologies that would transform the world.

During those heady days, nearly any startup with a "Dot-com" in its name could raise capital almost overnight in the race to go public, fueled by enthusiastic brokerages and institutional investors. The NASDAQ rose 86% in 1999 alone. But it wasn't just data centers attracting investment. Untold miles of fiber-optic networks were laid around the world, while networking and content delivery companies like Akamai soared nearly 400% on their first day of trading. The entire movement was driven by the promise of technological and economic transformation through the next generation of "killer apps"—remember that term?

At the time, Microsoft was considered a technology relic. Even mediocre talent was poached for exciting new ventures offering trendy offices, game rooms, on-site pubs, and lavish morale events. I was also heavily recruited with breathtaking equity packages designed to lure me onto the wave. I was slow to buy into the hype and probably missed the chance to hit the jackpot. But I was learning a tremendous amount in my role, and I was also traveling the world as part of Microsoft's Y2K task force (Yes, there were real threats, and we fixed them.)

What I saw behind the scenes during those global tours of data centers—and during job interviews—gave me pause. Huge IPO-funded data centers were filled with IPO-funded dot-com companies running services over IPO-funded fiber networks. I'm no economist, but it seemed that enormous amounts of money were changing hands to build all this infrastructure, while relatively few businesses appeared to have sustainable, profitable business models.

I decided to stay at Microsoft and sold all of my tech stocks—including what I could sell of my vested Microsoft options. That turned out to be a fortunate decision. Microsoft fell from roughly $120 to $30. Amazon lost about 90% of its market value, and companies like 360networks went from a $13 billion market capitalization to insolvency in just a few months. By October 2002, the NASDAQ had fallen roughly 77% from its peak. The party was over.

Of course, that wasn't the end of the story. While capital seemed to disappear, there are always people who can see "where the puck is going." The distressed physical assets of failed companies were auctioned off for pennies on the dollar. Companies like Digital Realty Trust acquired dozens of abandoned data centers, stripping and salvaging barely used generators, chillers, and transformers while holding onto the best facilities. They had the patience to wait for the market to mature.

Digital Realty eventually went public and today has a market capitalization approaching $70 billion, with more than 300 properties around the world. Today, it is riding the AI wave as well.

I eventually left Microsoft and joined a group that acquired 26 distressed data center sites. Together, we built another dozen facilities across Europe and Asia before the company was acquired by what is now Lumen. There are always people who know how to capitalize on failure.

Today, the scale and impact of AI investment dwarfs anything we saw during the dot-com boom. Yet the enthusiasm surrounding AI feels hauntingly familiar.

Try this experiment: Ask your AI engine of choice whether AI is profitable. The answers are revealing.

Yes, demand for energy, skilled construction and operations labor, semiconductors, and the raw materials needed to build AI infrastructure is outstripping supply. The build-out is contributing meaningfully to global GDP. But I still find myself wondering: when will AI itself become consistently and sustainably profitable? What is the "killer app" that ultimately justifies this unprecedented level of investment?

I realize profitability is only one dimension of our complex relationship with AI. But even if AI achieves the kind of singularity that Sam Altman envisions, economic reality still matters. Technologies don't always evolve around investment bubbles.  If the business fundamentals fail to materialize, we may be in for another rough ride.

 

[Note: To get daily delivery of this blog by email go to Https://petersage.substack.com. Subscribe. The blog is free and always will be.] 

Easy Sunday: Jon Ossoff is the real deal

Democrats have a plausible candidate for president.

I saw him audition in Portland yesterday. He sounds like a president. 

My wife and I attended a fundraising event for him.

The event had the usual elements of a higher-dollar fundraising event, plus some extras. Huge home, splendid hors 'd'oeuvres, valet parking, good sound system, a professional photographer who will send photos to guests. It sends a message that Ossoff is a top tier candidate.

It was a warm day, mid-high 70s. Ossoff wore this open-collared shirt. 

The event followed the usual format.  First, a brief pre-gathering where Ossoff met with the "hosts," i.e. the people who invited their networks of friends and made leadership contributions. Then 45 minutes of visits, handshakes, and photos with guests. Then the hostess took the microphone to acknowledge other event organizers and introduced the headline introducer, Senator Jeff Merkley. 

Merkley spoke for five minutes praising Ossoff's hard work and effectiveness in the senate.

Ossoff spoke for about 20 minutes. I have heard well over a hundred presidential campaign speeches. Ossoff's was at the highest quality, comparable to Pete Buttigieg, Bill Clinton, and Barack Obama, speaking within the genre of serious people presenting a story of America's condition and what needs to be done to address it. He sounds like a smart, professional, urban, or suburban politician; not "folksy good-ol-boy." Trump and Bernie Sanders both lead with grievance. Ossoff is more optimistic. He presents as can-do and forward-looking. There is a JFK "new generation of leaders" vibe to Ossoff.

Ossoff offers something Democrats desperately needed but lacked in a Biden presidency. He is articulate and persuasive. He didn't meander. He didn't check off  a list of policy proposals and proposed laws in the way that both Hillary Clinton and Elizabeth Warren did, which had the effect of making them come across as policy staffers, not leaders. He did not sound canned and over-practiced. 

He did not hide from trouble spots for Democratic politicians. He said that for Democrats to establish a "durable majority" in the senate, Democrats must diversity on some issues. Democrats need to win in red and purple states to get to 50-plus senators. He warned that we might not want to hear this, but Democrats were crazy to have drummed Joe Manchin out of the party. "Who else is going to win in West Virginia?" 

Ossoff is Jewish. An inaudible question from the audience asked him about Israel. Ossoff is generally aligned with the J Street Group, a pro-Israel, pro-peace group advocating both security for Israel and a negotiated two-state solution that condemns Israel's settlements in the West Bank. Ossoff set out what I perceived as a middle course: sharp condemnation of  Prime Minister Netanyahu but within the context of support for Israel continuing to survive and be recognized by Muslim countries of the region. It is possible that his position will satisfy no one among Democrats, and the division will break up the party. Alternatively, it may be that a liberal Jewish critic of Israel is the only person who can thread the needle and hold the party together. Parties are stitched together by rhetoric and personality. 

Here is how he presents himself in his official senate photo. Serious. Black suit, black tie. He is compensating for his youthful appearance.

He is 39. He is married and has a three-year-old and one-year-old. He married in 2017 to a woman who kept her birth name. He graduated from Georgetown University's School of Foreign Service; he earned a Masters degree from the London School of Economics.

An astute female guest at the event told me she thought Ossoff was exactly the right amount of good-looking to win in politics. "Ossoff is a nine," she said. "The perfect spot. Ten is good in Hollywood, not politics. Newsom is a ten, and he has too much hair and looks too slick, and comes across as untrustworthy. Ossoff is just right." 

She added that anyone from California will have a hard time winning a presidential election. "Georgia is a good purple state. Of course I would vote for Newsom over any Republican. I just hope I don't have that as my choice," she said.

Scenes from the event: short visits, photos.

Note: To get daily delivery of this blog by email go to Https://petersage.substack.com. Subscribe. The blog is free and always will be.]

Jon Ossoff is in Oregon today

I am attending a Jon Ossoff event in Portland, Oregon, later today.

He is running for re-election as a U.S. senator for Georgia.

He is also auditioning to be the Democratic nominee for president.

Here is what his campaign fundraising invitation looks like:

This is typical fundraiser setup: tiered donations; the exact address not reported until the day before the event; and sponsoring "hosts" that include well-known officeholders.

The invitation material focuses solely on him as a Georgia senator. That is smart and necessary. Of course he is preparing himself for a presidential run, but he is showing that he has his eye on the senate job he is running for, not the next one.

His campaign fundraising material presents himself as a person who can win: "Sen. Ossoff's pth to victory is clear and proven."

I will give a report on the event on Sunday or Monday.

Note: To get daily delivery of this blog by email go to Https://petersage.substack.com. Subscribe. The blog is free and always will be.]

Trump is the most consequential president in my lifetime.

He is changing how the Constitution works in practice.

The Trump administration funds energy projects in red states. It terminates funding for projects in blue states. 

This isn't my accusation. The administration admitted it. They stipulated to it in a court document.

The Department of Energy (DOE) admitted in court filings that it canceled nearly $8 billion in clean energy grants based on whether projects were located in in states that voted for Democrats in the 2024 election. The department denied it at first, but later admitted it. They stipulated:

DOE accepts that neither the inclusion of the ARCHES grant nor any other grants in the October notice tranche was based on any programmatic, statutory, cost-reduction, or performance-based factor.

DOE accepts that the inclusion of grants in the October notice tranche was based solely on the political identity of the grant recipient’s state, i.e., whether the recipient’s location and/or place of performance was in a Blue State or a non-Blue State.

The stipulation langugage comes on pages six and seven: DOE, Exhibit F. 

Trump is a uniquely consequential president, the most consequential of my lifetime. He is re-writing our Constitution, and by consenting to his practice our institutions, the political parties, the courts, and the public are ratifying them. The 2024 election was in effect a Constitutional Convention. A compliant Congress and Supreme Court are the ratifying votes. The power to act as Trump does was latent, unspoken, and officially denied -- until now, Trump's second term. A new order is in place as relates to the balance of power among the branches of government: A president can do anything he really wants to do. 

Trump complains. He pretends there are constraints because it serves his purpose. It is for show. He is like a magician clinking prop handcuffs together to prove to the audience that he is constrained. He complains the big bad Supreme Court may not let him run for a third term.

Trump can exercise vastly-expanded executive power because he is demonstrating that his purpose is not equal justice. It is the opposite, to normalize executive power as a tool for unequal power. That is what makes his power grab acceptable.

Trump rides a wave of backlash in American history. He identified and amplified an ebb tide in American thinking. Hey, America. Barack Hussein Obama is a Black Muslim from Kenya! 

Trump's anti-DEI, anti-woke, race-whistling, open-misogynist behavior is a partial restoration of former relationships between races and ethnicities, between men and women, and between the U.S. and other countries. It is also assertion that unequal power is the natural state of existence. He can grab pussies if he wants. We can invade Venezuela if we want. He can call Black people low-IQ and undeserving of promotion. He can pardon lawbreakers if they are on his side. Government is not about fairness. It is about securing domestic tranquility and security by empowering the powerful and disempowering insurgent people and institutions who would disturb traditional relationships.

As I wrote here on July 21, equal justice and the equality of all men are the stated goal our patriotic documents -- the Declaration of Independence, the Gettysburg Address, the 13th, 14th, and 15th Amendments, the Pledge of Allegiance -- but it has usually been a minority opinion. The Civil War brought it to a majority for a decade. The Cold War competition with the Soviets shamed us into it during the 1950s and 1960s, but conservative opinions in the U.S. have always been present. Equal justice is not the goal of a majority of  White, tradition-minded Americans who elected Trump. They want justice tilted in their favor. They have a grievance: They are being displaced by demographic change and the civil rights revolutions and Trump is a vehicle for resistance.

Trump is protecting and rewarding his team. Trump makes visible an executive power that was latent in the latitude given presidents. Trump put that power to work. They can be as dictatorial and arbitrary as they want if Congress lets them. 

The only constraint is the need to keep the support of one third of U.S. senators. And Trump has the capacity to reward or punish the states those senators represent. And he does.

[Note: To get daily delivery of this blog by email go to Https://petersage.substack.com. Subscribe. The blog is free and always will be.]

I don't give financial advice anymore, but. . .

I still watch the economy and stock markets.

Skip this post. 

I am probably wrong. 

Just because I have the willies doesn't mean there is anything to worry about.

I notice that there is a rush to build giant data centers.

The rush is so intense that the public has gotten spooked. Cities and states are banning them, even though they are theoretically a boon to a community because there are construction jobs building them, and then the fabulously valuable buildings sit there, nearly empty, paying property taxes but putting no demands for public services. 

But it doesn't work that way. Data centers want special low-ball property taxes, they are electricity hogs that sop up cheap legacy power generation, and they need water for cooling. People fear what they represent: a Frankenstein monster technology that steals our jobs and humiliates us by doing everything better than humans do.

But today's blog isn't about the politics of data centers. It is about the financing of them and the systemic risk they bring to our economy. They are being built to suit the most valuable and powerful businesses on Earth. Those companies represent America's economic and technical hegemony. Their capital investment has kept our economy growing. Their leaders have become allies with the president. They are too big to fail. Each of the competitors wants to be the first and biggest artificial intelligence company, the brand-dominant provider of a world-changing technology. You don't get that by waiting for profits from customer business. 

So what's my concern? Most data centers aren't being built by Amazon, Meta, Alphabet, Oracle, or Microsoft. They are the tenant for the facilities. They are the "smart money," the people with the best insight as to the growth of the technology. They have offloaded the risk by establishing Special Purpose Vehicles (SPVs) to create an accounting wall between themselves and the people who built the facilities. The smart money didn't want that risk. No problem. So-called "dumb money" did: private credit.

Maybe there is a good match between the pace of building those centers and the lease income from the hyperscalers using them. Being 76 years old makes me wary. Artificial intelligence users have moved from the experimentation stage to the shop-for-value stage. They realize that much of what they do can be accomplished on something produced by Chinese technology for one twentieth of the cost of AI work done by the frontier chips produced by NVIDIA and the American platforms. Data centers are multi-year facilities for a technology that changes by the month. 

A portfolio can get ugly very quickly. Oracle has lost half its value in the past two months. 

I expect Oracle and every other "hyperscale" company to survive. They are political partners with the American president.  The American taxpayer is a backstop for the very-well-connected. If someone must lose money in a crash-and-burn scenario of mis-allocated capital, the pension funds that funded those SPVs can take the hit. We saw this in the Great Financial Crisis. The well-connected get bonuses. The public absorbed the loss and suffered.

Don't mind me. I am old. Buy and hold. Everything will be OK.

In business and government, we are led by smart, prudent people who make careful, well-planned decisions. Right??

[Note: To get daily delivery of this blog by email go to Https://petersage.substack.com. Subscribe. The blog is free and always will be.]

The SAVE America Act is a bad bill -- but not because it hurts Democrats.

The SAVE America Act hurts Republicans.

It doesn't inconvenience educated, prosperous people (i.e., disproportionally Democrats). 

It discourages the occasional voter, the poor, the less educated, culturally traditional women, and the elderly (i.e., disproportionately Republicans).

For reasons of principle and what is good for public engagement in our democracy, Democrats should continue to oppose the SAVE America Act. But in the event that Trump muscles it into law, Democrats can enjoy one silver lining. It helps Democrats.

The SAVE America Act does not change the law that requires voters to be U.S. citizens. The law would change how to document citizenship. About 12 percent of registered voters do not have ready access to the types of documentation that satisfies the proposed law's conditions. This is about 28.4 million citizens.

A U.S. passport is the gold standard for being able to vote. Prosperous, educated people are more likely to have a current passport than are people with less interest or capacity to travel abroad. For better or worse, Democrats have become the party of prosperous, educated people. Forty-six percent of Americans have a passport; 56% of Democrats do, only 43% of Republicans.

Under the proposed SAVE America Act, there is an alternative to having a passport: a photo ID, plus a certified copy of a birth certificate with the person's full name, the name of at least one parent, the place of birth, signature of the authorizing agent, and the official seal. No photocopies allowed; a certified original only.  About 80 percent of Americans can meet this hurdle, and here Republicans slightly more than Democrats, 84 percent to 79 percent. 

The problem for Republicans is that the name on the birth certificate needs to match the name on the photo ID, and women who take their husband's surname on marriage have a new name. Republican wives are more likely to do that than Democratic wives. Twenty percent of Democratic women keep their birth name; only 10 percent of Republican women do. There is a strong skew toward women with graduate and professional degrees keeping their birth names; those women skew Democratic.

The SAVE America Act requires that the documents presented be current, which skews against the demographic of the elderly who allow their passports or driver's licenses to expire. My father did, about the time he turned 90. The oldest Americans skew Republican (though not my father.)

Theoretically, the various pluses and minuses balance out between the parties. Theory isn't reality. Regular, persistent voters tend to skew Democratic, which is why Democrats over-perform in special elections and mid-term elections. Trump's populist appeal has been underestimated in polls that measure "likely voters" because the MAGA base includes a disproportionate number of people who vote rarely or on impulse. These occasional voters are those least likely to jump through the hoops to have a proper, certified birth certificate or valid name-change documents.

Intuitively we might think that Hispanic voters would be the ones with the most difficulty in having all their paperwork together, since it might require a different document for many of them: naturalization papers. But their recognition that their citizenship might need to be demonstrated, not presumed, is why this group is the one most likely to have citizenship paperwork available. Hispanics partially drifted away from Democrats in 2024 but are once again a solidly Democratic-skewing group.

Bipartisan Policy Center

If the SAVE America Act passed, there would be a busy period of document-chasing and grumbling about the hassles of bureaucracy, and government making easy things hard, and Trump trying to suppress votes. That is a bad message for Republicans at the worst time and circumstance for the least interested people in putting up with the hassle.

Trump is chasing the chimera of millions of "illegals" voting. The irony is that his solution suppresses the legal vote of many of his own supporters. 

[Note: To get daily delivery of this blog by email go to Https://petersage.substack.com. Subscribe. The blog is free and always will be.]

Update and correction

Isabella Lee Tibbets tells me that she did NOT vote to endorse Clay Bearnson.  She said she voted no on endorsing both the Democratic election winner, Denise Krause, and Independent candidate Bearnson.

I was confident that I saw a green check mark on the Bearnson vote, but must have been mistaken. I am sorry for the error. She wrote:

Hi Peter,

Normally I have appreciated your blog but this one I have to correct. You are in fact wrong in saying that I voted to endorse Clay. I voted NO on all 3 candidate endorsements. Please do what’s right and correct that.

Thank you for the correction.  

The exhilarating thing about being a candidate for office is that one can decide for oneself the people and positions one supports and rejects. Isabella is exercising that right.

I have written multiple times in this blog that candidates are defined not by the general nice-nice things they say, e.g., listening to voters, calls for transparency, support for progress, and assertions that one would use common sense. Instead, they are understood by what and whom they oppose. Public opposition to someone or something creates boundaries and edges in an otherwise amorphous brand.

We now know something defined about Isabella Lee Tibbetts: She doesn't think fellow Democrats in the local party should endorse or support Denise Krause or Clay Bearnson. 

She may win votes from that position. She may get raised eyebrows of surprise and disapproval, especially from people who like those two candidates. That is why her position has credibility for defining her political brand. She was willing to pay a price for a position she took, the risk of losing support. 

It is her campaign. A candidate creates a brand by burning bridges, and she has done so. 

As Leslie Gore sang in 1963. "It's my party, and I'll cry if I want to, cry if I want to, cry if I want to." It is Isabella Lee Tibbetts' campaign. Laugh or cry, she laid down her marker.

https://www.leetibbetts.com