Today, Oppenheimer’s downgrade of AT&T from “Outperform” to “Perform” represents a structural re-evaluation of broadband economics, one that goes beyond a standard reaction to a single quarterly execution miss.

Stripping the company of its price target completely, analyst Timothy Horan argued that high-velocity low Earth orbit satellite constellations have shifted from a rural infill solution into an immediate, existential terrestrial bypass threat. The instant capital market reaction supported this anxiety, sending AT&T shares down over 4% for their sharpest single-day decline in months, dragging the equity nearly 20% below its recent year-to-date high.

The core of the analyst’s thesis relies on fast technological scaling and shifting subscriber dynamics, an outlook further accelerated by SpaceX’s upcoming 1.75 trillion IPO. Oppenheimer projects that LEO providers will capture more than 2 million subscribers annually, securing a 10% share of the total fixed broadband market by 2030.

The coverage maps got the attention. The control point did not.

The contest that matters for U.S. operators is not satellites against towers; it is the intelligence that decides which one serves the customer at any given moment.

The AI-Telecom Brief

Insight at the Intersection

SpaceX did not simply buy spectrum in May. It bought a position in the part of the network where the most valuable decision now gets made. When the FCC approved SpaceX’s purchase of roughly 65 MHz of nationwide spectrum from EchoStar on May 12, 2026, most coverage focused on satellites and dead zones. The more consequential shift sits one layer up, in the software that decides, second by second, whether a phone connects through a tower, a small cell, or a satellite beam.

Consider the timing. Two days after the FCC approval, AT&T, T-Mobile, and Verizon announced an agreement in principle to form a satellite direct-to-device joint venture, with the stated goals of pooled spectrum, common device standards, and a multi-vendor wholesale market. Read as a coverage move, it looks defensive. Read as a contest over decision rights, it looks like the incumbents protecting the one asset they cannot afford to lose.

Even if you put AI ethics and tech's support of fascism aside (which you shouldn't), the quality control at major tech companies has become an embarrassing and dangerous mess as they try to justify years of hype.

A law enforcement document obtained by The Intercept shows police scan social media looking for posts opposing AI data centers.

Americans speaking out against artificial intelligence data centers on social media are falling under police surveillance, a confidential law enforcement bulletin obtained by The Intercept reveals.

A fusion center in Philadelphia combed through spicy internet comments from AI critics and concluded there is a growing risk of physical violence against data centers from “domestic violent extremists,” ranging from white supremacists to anarchists.

“Domestic violent extremists (DVEs) are likely interested in targeting artificial intelligence (AI) data centers, posing a physical and cyber threat to infrastructure in the Philadelphia regional area,” the Delaware Valley Intelligence Center wrote in a December alert.

Grays Harbor Public Utility District (PUD), a wholesale open access telecom utility in Washington state, will soon enter phase four of an ambitious fiber expansion project that will bring affordable next-gen broadband access to rural residents written off by the monopolies that were supposed to serve them.

The PUD also says that Phase 4 of the PUD’s fiber internet expansion in south Elma, Porter and Cedarville will be reached later this Spring, bringing access to locals who have been waiting for years for faster, more reliable, and more affordable service.

If Alphabet's record-breaking $85 billion stock sale signals investor appetite for AI-related offerings, we can see that investors are ready to chow.

Google’s parent company had initially intended to sell a first tranche of $40 billion worth of various equity instruments — two different classes of shares, plus smaller “depositary shares” priced to be accessible to a broader range of investors. But the offering was so oversubscribed that it raised $45 billion instead, CEO Sundar Pichai said in a post on X on Monday. Among the buyers: Berkshire Hathaway, still known for its love of value investing, picked up $10 billion worth.

In this episode of Unbuffered, Chris is joined again by Joshua Edmonds, CEO of DigitalC, for an ongoing conversation about one of the most ambitious efforts in the country to address the digital divide.

AI data centres are energy-hungry and resource-intensive, requiring huge amounts of electricity generation, cooling systems, land and water supplies.

The United Nations has issued a stark warning, urging countries to confront the escalating environmental costs of artificial intelligence (AI) as its rapid expansion intensifies demands on energy, water, and carbon emissions.

AI data centres are notoriously resource-intensive, consuming vast quantities of electricity, water for cooling, and significant land.

A report from the Canada-based UN University Institute for Water, Environment and Health (UNU-INWEH) also highlighted the technology's footprint from chip production, critical minerals, and electronic waste.

Americans ain’t puttin’ up with these things no more. Welcome to Virginia, ground zero for data center defiance.

n mid-April, a week before Virginia voters narrowly passed new congressional maps in response to Republican gerrymandering, groups of landowners and land preservationists in Northern Virginia quietly won a state appeals court battle against a deep-pocketed consortium of developers. They had sued the county over failing to follow state regulations about posting public notices involving a data center project.

Between the redistricting vote, Virginia’s legislative budget impasse over data center taxation, and state and federal lawmakers caterwauling from Washington to Richmond and back again, it’s not surprising that a suburban county court case didn’t really penetrate the dystopian news cycle.

The Benton Institute for Broadband & Society reports...

On May 29, 2026, the Office of Management and Budget (OMB) proposes to revise the Guidance for Federal Financial Assistance. The proposals might be among the most consequential changes to federal grant administration in more than a decade, particularly in the breadth of its policy conditions.

OMB is proposing to rewrite the foundational rules that govern how nearly every federal grant dollar—including broadband, digital equity, research, and community development funds—is awarded, conditioned, and potentially terminated. 

Every entity that receives federal grants or cooperative agreements—states, local governments, Tribes, universities, nonprofits, hospitals, and for-profit organizations—could be impacted. 

Millions of households remain on the wrong side of the digital divide. This paper centers the lived experiences of those directly impacted by digital inequities -- and reveals what we can do about it.

Today, Public Knowledge, UnidosUS, and the National Digital Inclusion Alliance launched a new white paper, “The Blueprint for Equitable Digital Participation,” a report grounded in what low- and middle-income households told us about getting and keeping internet service. Right now, Congress, the Federal Communications Commission, the National Telecommunications and Information Administration, and courts are making decisions that will determine who can access affordable, reliable broadband. The results will impact the public’s ability to participate in civic discourse, unlock economic opportunities, and obtain critical healthcare and education services – in short, their ability to thrive in daily life.

However, we must be clear-eyed about where we stand as a country:

Some copywriter at Microsoft is getting a real stern talking to.

In an internal document obtained by 404 Media, the tech giant let slip that it wants to “make people addicted” to its new personal assistant AI agent, Scout — an alarming admission, given that AI companies have tirelessly fought against the criticism that they design their models to be as engaging as possible, to the point of being psychologically harmful and fueling mental health crises.

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

Bari Weiss’ sole purpose at CBS News is to decimate the storied network and its flagship program 60 Minutes — and replace them with Trump-friendly propaganda.

Free Press hosted a robust panel discussion featuring Chad Marlow of ACLU, Dr. Safiya Noble of UCLA Center for Resilience & Digital Justice, Memo Torres of L.A.Taco to launch our Solidarity Over Surveillance campaign.

Throughout our nation’s history, government officials have spied on marginalized communities and created a surveillance network designed to criminalize people of color, immigrants and protesters — and chill dissent. Now, private companies and data brokers are partnering with the government to harvest more of our personal information than ever before.

Corporate and government surveillance violates our constitutional rights — and policymakers must put a stop to it.

Plans to build a 2GW, 10,000-acre data center in Socorro, New Mexico, have been scrapped following widespread backlash to the proposal.

Back in March, data center developer Green Data floated the idea of developing a 10GW solar farm that would power a 2GW data center located on land owned by the New Mexico Institute of Mining and Technology.

If permitted, this would have been the largest data center in New Mexico.

But the President of New Mexico Tech, Michael Jackson, said in a statement on June 2 that they would not be proceeding with the project.

“During the initial phase of this process, we identified several material considerations that led both parties to mutually conclude that proceeding under the current framework is not the right path at this time,” said Jackson.

The California State Assembly recently voted 67-1 to strip telecom oversight authority away from the CPUC and shift it to a more easily lobbied state legislature – and an as-yet-undefined state broadband office.

The effort still has a long road before it’s formalized.

Assembly Constitutional Amendment 9, authored by Assemblymember Tasha Boerner, D-Encinitas, now moves on to the California Senate, where it needs to secure a two-thirds vote before appearing on a statewide ballot before California voters.

The proposal would remove the state constitutional requirement to define and regulate telecommunications as a public utility, something long supported by telecom giants. Boerner’s amendment (and companion bill AB 2289) gives lawmakers leeway to strip the CPUC of its telecom portfolio and hand it over to a newly created state broadband office by 2028.

Some tech observers think that the Palantir overlord sees the end times coming, but his real motivation is likely much more mundane and self-interested.

Historically, South America has proven irresistible to certain inhabitants of the northern hemisphere eager to escape the consequences of their terrible actions. Argentina was the favored destination for thousands of Nazis after the collapse of the somewhat-less-than-thousand-year Reich, including Adolf Eichmann and Josef Mengele; Klaus Barbie, meanwhile, ended up in Bolivia.

On a somewhat lighter and more British note, the escaped “Great Train Robber” Ronnie Biggs fled in 1970 to Brazil where he lived large for decades, even recording a couple of tracks with the Sex Pistols, including one in which he asked God to save “Martin Bormann and Nazis on the run / They wasn’t being wicked, God, that was their idea of fun.” (Bormann at the time was thought to be hiding in Argentina; he was in fact lying dead, as all Nazis should be, in Berlin.)

Now another terrible northerner seems to be readying his own ratline to Argentina: the tech-and-finance overlord Peter Thiel. 

Why Mimicking Silicon Valley Fails and How Telcos Can Weaponize Their Inherent Infrastructure Moat--

Techco is dead.The definition was always consulting fluff.

McKinsey defined Techco as an evolution into “platform-oriented innovators akin to today’s top tech players.” The financial mechanics prove this is a category error. Telcos and tech companies share nothing but the letter “T”. The inherent financial structure leaves no room for debate: telecom runs on 1% R&D; tech runs on 15% or more. We do not write proprietary operating systems or build global developer ecosystems. We manage heavy, long-lived physical infrastructure, power, and heating loads, using standardized vendor hardware.

Attempting to secure a software valuation by changing corporate vocabulary does not alter the depreciation cycle of a cell tower or the CapEx of a fiber trench. Techco is simply the newest headstone in the telecom graveyard, buried alongside WAP portals, Telco 2.0, DSP, and the Smartpipe.

This article is historical, fact-based breakdown of why attempting to replicate Silicon Valley’s model routinely destroys investor value, what we must learn from these multi-billion-dollar failures, and where the infrastructure must actually pivot to generate returns.

Something changed in the last week.

Shortly after Uber COO Andrew Macdonald said that it was “getting harder to justify” spending money on AI as it was “very hard to draw a line” from that spend to useful consumer features (after its CTO said Uber burned its entire annual token budget in four months), Axios’ Madison Mills reported that one company had accidentally spent $500 million in the space of a month on Anthropic’s models after failing to set spend limits. A few days later, Mills would report that other companies were now looking for ways to reduce their AI spend.

That’s because, as I’ve said before, nobody can actually measure the ROI of AI, or even create a standard measurement of the cost of a task thanks to the inevitable hallucination-prone nature of LLMs and the ever-growing list of different harnesses and “agentic” (sigh) interfaces. Every different prompt and project and interaction can go wrong in a way that is hard to predict or plan for other than having an eternal vigilance that the supposed “intelligence” doesn’t do something catastrophically stupid, because LLMs have no thoughts, consciousness or ability to learn outside of pre and post-training. 

If you can’t measure how good something is, how much it might cost, or what your return on investment might be, it’s fair to ask why you’re even paying for it in the first place.

For years, the media industry has been predicting the death of television. First, cable was expected to replace broadcast television. Then satellite disrupted cable. Streaming was supposed to eliminate both. Today, artificial intelligence is expected to reinvent advertising and content delivery yet again. While the technologies continue to evolve, I believe many people are focused on the wrong debate.

Television isn't dying. Distribution is evolving.

Consumers are still consuming enormous amounts of content every day. They still watch sports, news, entertainment, and live events. What has changed is not the demand for content itself, but the pathways through which that content reaches consumers. The traditional cable box has become an app. Satellite services have expanded into streaming ecosystems. Fiber networks, wireless networks, connected devices, and cloud-based platforms have created new ways to access the same content. The screen survived. The distribution model evolved.