Back in 2024 I helped to launch a new events and media series focused on the science of longevity. We called it DOC. Over the past two years, DOC has thrived, and a fascinating community of leaders in medicine, finance, policy, and entrepreneurship has flourished. This year we’re hosting our third annual DOC October 21–23 at the Fairmont Sonoma Mission Inn, with a focus on the intersection of medicine and AI.  

At the time we launched DOC, many of you asked me why I was involved – I’ll admit it seemed like a bit of a curveball. But beginning with the Internet Summit back in 1998 and continuing through The Standard’s rooftops, Web 2, Foursquare, Signal, and NewCo Shift, I’ve always loved curating an influential group of people in the context of a larger, high-impact narrative. It’s journalism in a live setting, and if you get it right, it can catalyze a nascent industry and drive it forward. 

Longevity science is exactly that kind of nascent industry today. Not only is it driven by many of the same forces as the tech and internet sectors – particularly data-driven innovation like AI – it’s also enduring almost exactly the same hype and investment cycles. The category is already valued at nearly $7 trillion dollars, and is set to grow by $3 trillion in the next few years. As it does, the market has been inundated with what DOC’s co-founder Dr. Jordan Shlain calls “bad science with good marketing” – wellness influencers and opportunists making grandiose claims based on questionable scientific evidence. 

We launched DOC to cut through that noise and identify what’s real, impactful, and scalable right now. We identify and feature a Faculty of researchers, entrepreneurs, and policy leaders who believe in the power and practice of quality science, and we fill the room with the kind of folks who can make a direct impact on the field. And as is the case with any great gathering, the true breakthroughs aren’t delivered from the keynote stage. They happen in the spontaneous side conversations between the individuals driving real change.

We’ve reached a critical inflection point where longevity science and clinical AI are transitioning into practice much faster than our society can absorb their impacts. Real advancement doesn’t occur in isolation within a single lab or fund; it sparks from unexpected intersections. DOC is explicitly designed to foster these collisions. The 2026 faculty brings together pioneers across fields: leaders like George Yancopoulos (Regeneron), Alicia Jackson (ARPA-H), and Laura Esserman (UCSF) representing the science; Max Hodak (Science Corporation) and Travis Zack (OpenEvidence) at the leading edge of BCI, AI, and entrepreneurship; and key investors like Divesh Makan (ICONIQ Capital). For three immersive days, they share a single room for genuine dialogue rather than scripted speeches or conventional panels.

Notice the mix. Scientists who publish. Clinicians who treat. Builders who ship. Investors who bet. Put them in one room for three immersive days and a claim about artificial intelligence in the clinic gets examined by someone who codes, someone who diagnoses, and someone who pays. That triangulation is how a field grows up.

If you have already received an invitation to DOC 2026, consider this the sign to secure your place. If you belong in this room but haven’t been invited, you can submit a request at doc.health. Capacity is strictly limited, and we are close to sold out. I hope to see you there!

—

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.

Earlier this week I posted something of a coda for the open web, but today news broke of a small victory for “how things used to be.”

The backstory: For decades, Google has been the beating heart of the Internet’s ecosystem, pumping the oxygenated lifeblood of traffic across tens of millions of revenue-starved websites. That’s all changed over the past few years as Google has erected a massive set of AI-driven walls around its core search service. As I said in “Google Encloses the Web,” Google no longer views the web as a destination, but rather as “raw material” for the company’s core product, which, like so many of its peers, is attracting and retaining human attention it can sell to advertisers and, via Gemini subscriptions, to businesses and consumers.

Regardless of Google’s recent moves, there remains a cottage industry of small firms that earn their keep by helping businesses make sense of the search ecosystem. These firms build databases of Google results, then sell that data to a large swath of the Internet economy. The practice has always lived somewhat in the gray – Google’s internal policies forbid the bulk gathering of Google’s own search results. Like Google its peers Amazon, Meta, Netflix, and many others, Google expends herculean resources to defend those policies. And until recently, the question of whether Google or its data-scraping adversaries have legal standing has not been put to the test.

SerpApi is one such adversary. It has built a business by scraping Google’s results and packaging them for clients interested in understanding the trillion-dollar ecosystem that Google commands. On its site, SerpApi claims Uber, KPMG, Shopify, and NVDIA as customers, among many others. Late last year, Google decided it had enough, and sued SerpApi for “running an “unlawful” operation that bypasses Google’s security measures,” essentially selling “a back door to Google’s proprietary search engine.”

At issue is whether or not search results – and the queries that drive them – constitute public data, or if Google can claim proprietary ownership. SerpApi claims it is “backed by the First Amendment” and the Fair Use Doctrine, but until now, its arguments have been mostly academic.

Yesterday, a U.S. District Judge handed SerpApi a clear win, dismissing Google’s suit and claiming the search giant did not have standing. The details of the ruling are worth considering: Google claimed SerpApi was bypassing Google’s anti-bot security protocols, in violation of the DMCA, a 1996 law that, among other things, makes it illegal to circumvent technologies that protect copyrighted material.

The judge determined that while SerpApi most likely did use circumvention to scrape copyrighted information – it’d be impossible not to, given the content of Google search results – Google could not claim it was acting on the actual copyright holders’ behalf by enforcing its anti-scraping policies. Put another way, Google has no standing to claim the anti-circumvention protections of the DMCA, because Google is, in essence, the biggest scraper of them all.

Why does this matter? It may not, given the size and scope of Google’s legal team and its lobbying muscle in DC. But it is a fascinating precedent given the deteriorating relationship between Google and actual copyright holders like Dow Jones, Axel Springer, and CNN. In a recent Wall Street Journal article, all three (plus many other publishers) were reported to be considering abandoning Google’s index altogether, in large part because of Google’s decision to enclose the web. “Turning them off and blocking them entirely is 100% on the table,” People Inc. Chief Executive Neil Vogel told the Journal.

As the web shifts from open to closed, from intelligence at the edges (us), to intelligence as a service (the AI providers), it’s worth paying attention to the little stories that might just point to a new architecture of control on the web. The SerpApi decision is just such a story.

—

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.

Every so often the tech world appropriates a perfectly pedestrian word and imbues it with meaning well beyond its settled definition. For centuries “computer” was an obscure term referencing an obscure craft: people who performed mathematical calculations. And prior to the mid 1990s, anyone talking about a “web” was probably discussing spiders.

“Token” has broken out as technology’s latest etymological adaptation. The word has a long history in the tech world, and even richer origins in early English and Germanic languages. But as AI has ascended to primacy in tech,  “token” has become a way for executives, financiers, and journalists to grasp the maddeningly opaque workings of an industry that seems to be swallowing our economy whole.

Until recently, anyone in the tech industry obsessing over “tokens” would have been talking about crypto, that much-maligned domain of hustlers, crooks, and con men. While crypto is a punchline today, just five years ago blockchains, “NFTs,” and “initial coin offerings” were all the rage. Venture capitalists like a16 were raising billion-dollar “Web 3” funds. Tokens lay at the center of this resurgent crypto world: these “programmable digital assets” offered a new approach to financing, governance, ownership, and more. The world was going to change, big time, and we had tokenization to thank for it.

Crypto never lived up to its own hype, and for the most part the sector has been marginalized by the Valley’s Next Big Thing – generative AI. I’ve always thought there was a natural symbiosis between what crypto is good at and what AI needs. Regardless of whether the two merge, it’s clear that right now, we really, really need a way to price the value of AI in society. And to get that work done, we’ve once again turned to the word token as a container for that work.

What’s the Business (Model)?

I’ll begin by stating the obvious: For better or worse, investments in AI related sectors has ballooned to bubble-like proportions. Major tech platforms now carry $1.65 trillion in AI-related debt. Just this year, nearly $1 trillion will be invested in data centers alone. In the first quarter of 2026, 75 percent of all VC dollars went into just five AI companies – OpenAI, Anthropic, x.AI, Waymo and Databricks. AI-related firms hold a market capitalization of more than $27 trillion – more than a third of the entire US stock market. Dozens of market analysts, lead economists at major banks, hedge fund managers, and even AI cheerleader-in-chief Sam Altman have warned that we’ve over-invested in AI.

With trillions at stake, a question naturally arises around return on investment. When will AI companies (and their clients) start producing profits that justify those enormous outlays? And most importantly, how will they do it?

I’ve been posing exactly that question to dozens of leaders in tech over the past few months, and so far I’ve found credible answers elusive. Almost everyone agrees that whatever models we come up with now, things are will change dramatically over the next few years. Some believe companies like Anthropic and OpenAI will end up becoming trillion-dollar cashflow machines, mostly through conquest of massive markets like software, legal and health services, and retail. Others believe those same companies will become commodity providers to a more distributed economic revolution that drives unprecedented growth across all market sectors. And more than a few have predicted a crash: a necessary correction jolting us into whatever new structures we’ll need to create to accommodate “intelligence as a service” in our society.

That last answer feels the most likely to me – it rhymes with society’s adaptation to the Internet in the years following the dot.com bust. However all that future growth ends up happening, “tokens” have become the consensus mechanism for how we’ll count the money along the way. You’ve probably heard the term quite a bit in the past few months – breathless stories of “tokenomics” and “tokenmaxxing,” which has cost organizations from Uber to the US Army hundreds of millions of dollars.

So what are tokens? According to my pal Google, in the context of AI, they are “the fundamental units of text that an AI model reads and generates. In English, 1 token is roughly equivalent to 4 characters or about 1/4th of a word. AI providers charge for usage on a per-token basis, categorized into distinct, billable stages.”

Put another way, tokens stand in for “delivery of value” by an AI service provider. You prompt AI, it burns a certain number of tokens to deliver you a response, and you get charged on a per token basis. This simple usage model is driven by the very real costs discussed above: Data centers, Nvidia chips, electricity, and billion-dollar engineers.

For the past year or so, corporate America drank from the token hose like there was no tomorrow – and the AI labs were happy to subsidize their true costs so as to capture new customers. But in the last few months, the bill has come due, and no one is sure how to split the check. All these economic realities are forcing the nascent market into necessary and predictable rationalization: What is the true value of AI? Might there be more efficient ways to acquire it than simply giving everyone a Claude Code account, then holding your nose when the invoice comes?

Again, History 

Cast your mind back 75 years, when a remarkable new technology emerged offering “intelligence as a service.” The newfangled mainframe computer could run payrolls, calculate missile trajectories, and tabulate massive datasets like the US Census. But mainframes were unwieldy and expensive machines that took months to set up. It made little economic sense for corporations to have one of their own. Instead, early tech firms like IBM created “service bureaus” which allowed companies to “time share” mainframe computing resources. The bureaus metered each client’s usage based on how much time and CPU processing cycles they consumed.

Over the next few decades computing migrated from the centralized mainframe to the distributed personal computer, and the concept of a centralized compute resource fell dormant as corporations took their computing budget in house. But then came the Web, with its insatiable demand for domain hosting, bandwidth, and storage. Once again it made sense to “rent” your compute from a centralized “cloud” based on a usage model (a model which, by the way, was orders of magnitude cheaper and far easier to use than its predecessor). And it’s that very model – a centralized, metered service – that dominates AI’s current economic playbook. Only this time, the element being metered isn’t processing time or storage capacity, it’s token usage.

Will that model last, or will the AI industry see an evolution similar to what  happened with computers? It’s hard to say, but one thing is certain: Intelligence as a service has to get far less expensive if it’s going to become ubiquitous. And plenty of trends seem to point in that direction. Far cheaper Chinese AI models now account for 41% of open-source downloads on Hugging Face, according to one report. The market is speaking: Paying for AI is fine, but overpaying is not.

As Enrique Dans, a keen observer of AI’s impact on business, puts it: “a growing portion of what was previously accounted for as human labor hours is now expressed in terms of computational consumption, iterations, context, agents, and tool calls.” The accounting method? Tokens. “The company that understands this transition will learn to manage tokens as a scarce resource,” Dans advises.

Dans is right, and in that insight lies another: When tokens become currency, everyone can compete on the cost of that currency. Just as it did with the rise of the cloud, the architecture of the internet will once again shift, with potentially tectonic implications. Only this time, the internet isn’t just an interesting sandbox where (mostly American) businesses are trying to figure out how to have a dot.com presence. This time, the whole shooting match is in play.

—

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.

I’ve always loved writing, and took to this form – the “thinking out loud” version of blogging – more than 20 years ago. I started Searchblog after I sold my first book, which focused on the impact of search on business and culture.  I decided I’d chronicle my reporting and build my voice by posting as often as I could, and the experience proved to be a watershed moment in my career. I built an audience that numbered in the hundreds of thousands, noticed scores of others doing the same, then launched a new company (Federated Media) based on insights from the experience.

The early 2000s marked the apex of what is now called “the open web,” and today two stories may as well be considered obituaries for the era.  The first, by Kate Conger at the NYT, lays out the mechanism for the open web’s demise: “Google Is Building an A.I. Fence Around the Internet It Once Championed.” The second brings news of a $1.5 billion settlement in a massive copyright infringement lawsuit targeting the AI industry. As an author, I’m a party to that suit – and at some point in the next year, a check from Anthropic will wend its way through the mail to me. A coda, of sorts, for an era already passed.

As Conger details in her piece, Google search once drove most of the internet’s traffic. It created a positive feedback loop between high quality content, often found on early blogs, and burgeoning consumer usage of search engines. When this site launched, there were relatively few blogs, and because we reliably linked to quality sites, Google’s search index  rewarded us with high “pagerank” scores, which meant our posts ranked at the top of many high-traffic queries. If you were looking for coverage of business, you’d likely be routed to Business Insider. Internet culture? Google directed you to Boing Boing. And news about Google itself? Well, Google itself would often direct you to Searchblog.

Hell, for a minute there, my site was the first link that appeared after typing “John” into Google. That was heady, though I certainly knew it couldn’t last.

The rise of AI has only formalized what has been a steady decline in the open web. Federated Media’s revenues peaked in 2012, and we sold it to a larger media company two years later. The rise of social media, and then apps, made it abundantly clear that the world was headed toward a landscape of walled gardens and siloed data. Yes, there’s still Wikipedia and Substack, but these are exceptions proving the rule: the open web didn’t usher in a more perfect world of “small pieces loosely joined.” Instead, its bones fertilize the soil for whatever strange crop AI will reap.

So I’m feeling a bit wistful as I consider a check from Anthropic that I doubt I’ll cash. I’ve no idea how they’ll determine payouts, but suffice to say the number won’t prove life-changing. Instead, I think I’ll frame and hang it right next to my first editions of Wired and The Industry Standard. An artifact testifying to that brief and glorious moment when the web was young, free, and unencumbered.

—

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.

At the time, I wondered out loud: If “chat is dead,” what happens to OpenAI’s promises of $100 billion in advertising revenue by 2030? Given the “surface” for that revenue was supposed to be chat, it seems contradictory to fold it into a paid subscription product focused on white collar workers.

A version of the super app (called “ChatGPT Work”) launched this week and early reviews are mixed: chat has been demoted, the UI is messy and confusing, but the “work” portion of the app is cheaper than Claude and quite good overall.  I’m sure the app will quickly address those UI issues, but I’m still left scratching my head: What about the ads? If I were considering investing in OpenAI’s ads, I’d certainly want an answer to that question.

Apparently ChatGPT remains a pure chat application on mobile, so perhaps that’s where OpenAI’s ad business will focus.  But OpenAI risks alienating marketers with its confusing pivots and lack of clear narrative for its ads product. Marketers need to know that a new go-to-market channel will be consistent, stable, and here for the long term. Otherwise, they’ll wait and see – and that $100 billion revenue goal will be increasingly difficult to hit.

—

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.

It seems ridiculous now, but when I was a cub reporter covering the tech industry, our publication had an entire beat focused on word processing – and it was one of the most sought after assignments in the newsroom. We had another reporter who focused on spreadsheets. Because I was the new kid, I covered databases – databases! – and other reporters covered desktop publishing software and utilities. At just one publication, that’s a half a dozen full-time reporters tracking and analyzing applications that have mostly faded into the background noise of today’s tech industry. At the time (late 1980s), there were at least a dozen outlets doing the same thing.

Think about that – nearly one hundred journalists covering workplace software!

But if you put that era in context, it certainly rhymes with today’s obsession with AI. Think of “word processing” as “ChatGPT” and “spreadsheets” as “Claude Code,” and you have a pretty good sense of what it felt like to be covering tech during its initial rise. An entirely new class of products were revolutionizing the workplace, and it seemed like everything was at stake.

Software was the story of the PC age.  Operating systems had consolidated into two players: Windows and Macintosh, with the Mac very much David to Microsoft’s Goliath. But on top of those two platforms, hundreds of software companies competed in what felt like a heady free for all. Sure, there was Microsoft Word, but there was also WordPerfect, MacWrite, WordStar, MultiMate, and dozens of tiny competitors like Perfect Writer (I worked there for a hot minute). In spreadsheets, there was Lotus123, Excel, VisiCalc, Borland Quattro, SuperCalc, and again, another few dozen smaller competitors. The same was true for databases, desktop publishing, and utilities, where the competitors numbered in the thousands.

By the mid 1990s, Apple had lost both the hardware and the software races, and faced financial disaster as a result. Just as the web platform shift was building, Microsoft came to Apple’s rescue with a $150 million investment that guaranteed Apple users would have access to Microsoft applications for at least five years.

Why am I going on about all this history? Because so many in our industry seem to have forgotten it. Steve Jobs’ deal with Bill Gates was all anyone could talk about in 1997 – and yet a few years later, the world wide web obviated the software industry as we knew it. My point? Platform shifts are rarely understood in the moment, even by those with the most at stake.

Last week I wrote about Palantir CEO Alex Karp and Microsoft CEO Satya Nadella, both of whom are pushing a narrative about how OpenAI and Anthropic are bad faith actors hell-bent on “stealing your alpha.” Thinking about their complaints in the framework of tech history, they both end up sounding an awful lot like the leaders of legacy software companies protesting the ascendance of upstarts like …. Microsoft and Apple. Bill Gates famously “borrowed” his DOS operating system from a legacy company called Digital Research, which was the fourth largest software firm in the world at the time. And Steve Jobs famously “borrowed” the tech behind the Mac from Xerox’s PARC lab. Xerox was one of the largest corporations in the world at the time.

The tech industry maintains a tangle of contradictory maxims and mythologies, but principal among them is the idea that at any given time, an upstart will break through and “change everything.” It’s that conviction that fueled Andy Grove’s famous statement claiming “only the paranoid survive.” And it’s the foundation of the Valley’s venture investing community, which made its fortune betting on “disruptors” like Netscape, Google, Tesla, Salesforce, Uber, and now Anthropic and OpenAI.

The last few platforms shifts – mobile, social, web, PC – were built over decades as infrastructure, hardware, software and consumer habits crystallized into somewhat predictable industry sectors. But the AI shift is happening over a few years, with every major tech player running headlong at the opportunity without a clear sense of how things are going to shake out. They must both protect their flanks while embracing a mentality of disruption – a contradiction that’s beginning to bite.

Despite the tech industry’s idolatry of disruption, Palantir and Microsoft are warning their customers to avoid frontier models altogether, and Apple is closing the app store to AI startups and  suing OpenAI. But if these two giants are going to thrive in a post-AI world, they’ll have to find a way to embrace the chaos and build platforms that are true to their roots as original tech disruptors. While they fight rearguard actions, thousands of startups are trying to build the next application layer for AI – the equivalent of spreadsheets and word processors back in the PC era. Will it be coding or consumer agents? Or will it be an entirely new class of products and services from a company or companies we’ve not yet heard of?

If the past is any prelude, the latter will certainly be the case. But this time around, the incumbents aren’t just industry leaders. They’re the most powerful organizations in the history of humanity. How they respond to this moment will define more than the course of an industry – it will shape civilization itself.

—

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.

Have you seen Alex Karp’s rant on CNBC? I fielded this question about a dozen times over the past week. If you’ve not yet watched, it’s well worth your time. In a conspiratorial diatribe against the evils of monopoly-crazed frontier labs, the founder of defense-tech giant Palantir asserts that OpenAI, Anthropic, and Google are ruining our economy by “stealing our alpha” (more on that in a second). And because he’s a founding member of the Valley’s newly emboldened MAGA elite, Karp manages to toss in a few unhinged political dog whistles, implicating both UC Berkeley and “wealth taxes” as complicit in the downfall of civilization as we know it.

Despite obviously talking his own book – Palantir sells an application layer that sits above frontier models and ensures clients will not have their “alpha” stolen – Karp makes a point that has resonated deeply across not only tech, but the entire economy. “He’s not wrong…” is how several of the emails I’ve received have begun. “He’s got a point,” they continue, “even if he is kind of toxic.”

Karp’s point has been made before. Microsoft CEO Satya Nadella laid out the same argument last month. In “A Frontier Ecosystem Without an Ecosystem Is Not Stable,” Nadella issued a call to arms to the business community: “The last thing any of us want,” he wrote, “is a world where every company across every sector is ceding value to a few models that eat everything they see.”

There’s a lot to unpack here, but let’s focus on the concept of “stealing my alpha.” What the hell does that even mean? The term is borrowed from the financial sector, where “alpha” represents the excess return or competitive edge a particular investment yields over what the market might usually return (this is where the leading finance publication Seeking Alpha got its name). In Karp’s usage, “alpha” stands in for any firm’s unique right to earn profits in any given market. According to both Karp and Nadella, AI models have unfettered access to the core data and insights driving a company’s ability to make money, and they will use those plundered assets to steal the company’s customers (the first at-scale indication of this theory was the “SaaSpocolypse” in the software sector earlier this Spring.)

Simply put, Karp and Nadella are arguing that a novel tech platform – AI models, in this case – are disintermediating entire sectors of our economy and positioning themselves as voracious monopolies who will ultimately control … well, everything.

To this notion I ask: Where have you been these past 25 years? The rise of Microsoft, Google, Facebook, Amazon, and Netflix was powered almost entirely by those firms “stealing the alpha” of entire sectors of our economy. Media, advertising, communications, retail, software applications like browsing, calendaring, and contacts – all swallowed by alpha-seeking platforms executing what Ben Thompson famously called “aggregation theory.”

AI models are simply running the most successful play in the tech industry’s playbook. Now, all of a sudden, we have a problem?! What’s new here?

To answer my own question, I think what’s new is that we finally have a constituency with the power to oppose the wholesale appropriation of data that’s been the mainstay of the tech industry these past few decades. Unlike consumers, who willingly ceded control of their data for the convenience of tech services, corporations are terrified of sharing information that might expose their strategies, competitive edge, or weaknesses. Karp and Nadella know this, and are exploiting that fear to build their own businesses. “Working directly with OpenAI or Anthropic is madness,” they’re saying. “Work with us instead. We’re on your side.”

I’ve yet to see a formal response from the AI model companies to this charge, but I’m sure one is coming. And as much as I’m enjoying watching the Karps of the world argue against their previously held dictums – wasn’t it Peter Thiel who argued that the natural state of any tech company was to be a monopoly?- I’m far more interested in how this might play out for the rest of us – we customers, consumers, and individuals whose data has always been viewed as a free resource to be exploited. In Will AI Ever Become a True Platform? Not Without Us, I wrote “Nadella’s essay is a declaration of AI independence for companies around the world. If only we’d extend that thinking to the individual, we’d have a real revolution on our hands.”

Perhaps once corporate American puts its foot down and demands agency over how its “alpha” is exploited, we’ll collectively demand the same as consumers. I want my alpha and my omega, and I want it now!

—

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.

One of the best ways to find signal in the noise of a major technological shift is through the time-tested journalistic tactic of following the money. Back in January I predicted that in B2B markets, AI Can’t Cost This Much. I argued that “pricing the future based on the cost of the present is a losing bet … However many trillions will be spent on AI over the course of the next decade, one thing will remain constant: There are huge opportunities to be exploited in finding more efficient and less expensive ways to deliver technology to both consumers and businesses.”

Nearly three years before writing that post, I made another prediction, this time in consumer markets: It would soon be normal for consumers to pay as much as $200 a month for AI services. Combined, we pay more than that for mobile phone, cable, and internet services, after all.

Have both of my predictions come true? While it’s certainly not “normal,” millions of us are in fact paying $100 a month (or more) for ChatGPT Pro, Claude Max, and/or Gemini. And tens of millions of us are paying at least $20 for basic AI subscriptions. It’s absolutely in the interests of these AI providers to upsell us – think of the cost of your very first cable bill compared with the last one you paid. I quit cable a few years ago, but before I did, my bill averaged more than $200 a month.

So what about this year’s prediction, that AI costs will come down significantly in the enterprise/business world? The answer to this one feels far more certain. Here are three headlines from my morning reading today:

Microsoft Replaces OpenAI, Anthropic With Own AI in Some Apps (Bloomberg)

AI Giants Are Handing Out Tons of Free Computing Power to Grab Startup Share (WSJ)

China’s Answer to AI Sticker Shock (The Atlantic)

Taken together, these three stories paint a picture of a market acting rationally in the face of downward pricing pressure. Microsoft is cutting its OpEx by replacing expensive frontier models with cheaper versions it built in-house. Those frontier models are attempting to secure replacements for that lost revenue by luring startups into using their services through steep discounts and introductory pricing offers. Meanwhile, Chinese models built with far less expensive inputs (DeepSeek, Z.ai) are starting to take over the enterprise market – recent data suggests nearly 40 percent of US corporate AI use now depends on these far cheaper models.

The tech industry spent the lion’s share of the past year figuring out how to pay for investments in data centers and chip foundries with fixed cost  assumptions that may no longer pencil out. With the AI marketplace in the process of a significant realignment, it certainly feels like there’s a fair bit of economic headwind ahead. There’s also plenty of opportunity – if we let the free market do its work.

—

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.

“Our lives have become dematerialized.”

That phrase – from author Ian Bogost in a short piece in Wired – struck at the center of something I’ve been trying to verbalize for years. Bogost argues that technologies of convenience and efficiency have destroyed our connection to the physical world. His primary example is the automobile – first the automatic transmission replaced our physical connection to the road, then the EV made gears obsolete. Now driving itself is retreating into a digital interface – Waymo, Uber, and Tesla are seeing to that.

I’ve long argued that whatever can be digitized, will be digitized. In early eras of the digital revolution, that meant office work – typewriters, calendars, and rolodexes tossed into the digital wood-chipper and smoothed into “software” painted across glimmering green cathode ray tubes. Later, the rise of search and the web augured the end of materially bound information – vinyl albums, bulky yellow pages, printed magazines, books and newspapers. Then apps swallowed the rest of our interactions with the world – from dating to banking to meetings and getting a reservation at a restaurant.

I was an early and avid enthusiast of this shift – I imagined a more perfect world of knowledge and experience, freed of the constraints of the physical world. But lately I’ve begun to question this philosophy.

Software has indeed eaten the world, but I believe we’ve not entirely considered the consequences of the meal. The loss of direct interactions with other human beings is certainly the most obvious downside. What happens to us when physicality becomes optional, and we can engage with others through the mediating glow of smoothed glass and pixels? We’ve two decades of data on that question already, and if you think it points only to progress, well, I respectfully disagree.

For some, this transition is not only positive, it’s an inevitable step toward humanity’s destiny as an immaterial species. And what’s more incorporeal than God? “Silicon Valley has a God complex,” writes Reid Hoffman, one of tech’s more thoughtful observers. He describes the Valley religion as containing many “denominations,” including Humanists and Missionaries, who I’d argue are wary of dematerialism, and Narcissists and Accelerationists, who I’d argue are most certainly not. If technology is a lever, Hoffman writes, “The Narcissists believe the lever exists to elongate their own shadow. The Accelerationists believe the lever must be pulled without hesitation or respite.”

“No other animal has rewritten its own cognitive architecture through what it builds,” Hoffman continues before concluding that we should not fear technology’s progress, but rather embrace a Humanist approach to its integration in society. I agree with him, but I’m increasingly convinced the Humanists are losing the argument.

There’s always been a particularly virulent strain of “narcissistic accelerationism” in tech, but for most of the past 40 years, it’s been muted. With the rise of AI, the political philosophy of tech has shifted toward raw power and unobstructed capitalism. And it’s that combination – the power of actors like Musk, Karp, Theil, Andreessen, and Trump fused with the seductive whisper of a God-like artificial intelligence – that has been worried we’ve lost the plot. If we’re on an inevitable journey toward immortality, if we are destined to be freed of this mortal coil, well, what’s the point of caring about the problems of mere mortals suffering in the physical world? It’s unfortunate, of course, that so many must suffer now, but in a generation or two, nirvana awaits.

Right?!

–

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.

It seems almost quaint now, but there was once a version of the Internet ruled by influencers that were honest, high integrity folks who genuinely cared about the impact of their words.

Perhaps the most thoughtful of these was Om Malik, who passed away last week at the age of 59. Om was a colleague and a contemporary, and his site GigaOm was one of the earliest and most important partners at Federated Media, an early “record label” for bloggers that I launched in 2005.

I had read Om’s site for more than four years before reaching out to him and asking if he’d join our “federation” of early tech bloggers. He agreed, joining pioneering sites like Boing Boing, TechCrunch, AVC, and yep, this very site (hard to believe, but in the early years, Searchblog had more than 250,000 readers a month).

I think it’s fair to say that Federated was one of the very first platforms for “Internet influencers” – and certainly the first to wade into the somewhat murky waters of how to get quality authors paid through marketing partnerships. This all emerged in the early 2000s – a full quarter century before we got…whatever this is.

Om was a crucial thought partner for me as I attempted to square the maddening circle of what I called “conversational marketing,” which I defined as a new approach to advertising that acknowledged media had become a conversation, not a dictation, and for marketers to succeed, they had to partner with new forms of media driven by direct engagement with both publishers and their audiences.

I argued that this new era of conversational media was driven in large part by the rise of search, itself a new kind of “conversational interface” that had rapidly become the core driver of value on the Internet. (If you want to review a presentation about all this, here’s a link to a PDF of a talk I gave at Berkeley back in 2007).

Om challenged my thinking on any number of fronts, but crucially, he also agreed to work with me as we experimented with new approaches to publishing partnerships. One such experiment was a deal with Microsoft, which was looking to prove its relevance in the burgeoning world of blogs. The resulting campaign, called “People Ready,” sparked a controversy that still carries lessons for today. (It also led to new FCC regulations related to influencers – regulations that, alas, have been largely ignored.)

The campaign invited publishers like Om (and myself, as well as many others) to comment on Microsoft’s marketing messaging. If I recall correctly, the prompt was “What makes your business ‘people ready’?” Our responses were then integrated into advertisements that ran across the Federated network.

We’d been running similar campaigns for more than a year, and at first, no one noticed. But two months in, Valley gossip site ValleyWag decided to call us out, and all hell broke loose.

It was a shitstorm, and Om and I were at the center of it. Om quickly decided his integrity was more important than the ad dollars he was making with Federated, and pulled out of the campaign. Others, including TechCrunch and Fred Wilson at AVC, defended it. For about a week, “People Ready” was all anybody in the blog world could talk about. And while I probably aged a few years in that week, I also grew as both a CEO and as a writer.

The People Ready campaign is now a forgotten piece of Internet history, but as evidenced by the outpouring of tributes and grief, Om will be remembered well and forever. Long after Federated was sold and we both moved on, Om and I stayed in touch, swapping notes on the industry and comparing possible new investments. He was ever the true gentleman, and his writing on the state of our industry stayed potent until the very end.

Rest in peace, good sir. You are deeply missed.

–

You can follow whatever I’m doing next by signing up for my site newsletter here. Thanks for reading.