Jim's Vacation Announcement

Hey, it's Jim. Just wanted to remind you that I'm off this week. I will be back next Monday, but with the two holidays and a lot of stuff going on, I'm taking the week off. If you have been one of those great people who want to get a pre-copy of my book, then I'll be writing to you this week.

I'll be getting some of the things organized. and if you still are interested but you haven't quite written in, you can still go to the Contact Us form on the website and leave me a note with your email address and I'll put you on the list for the pre-reading of my new book, The Compassion Virus. I heard that some of you may not have gotten Friday's episode because of a problem with our MSP, but I'm going to rebroadcast that for Monday. If you've heard it, just skip on.

And if not, then some of the stories are still pretty good. And probably this weekend on the holidays, I will rerun something, probably an old Project Synapse or something that I thought was particularly appropriate. In any event, have a nice week away from me, and I'll be back next Monday. This is Hashtag Trending for Friday, June the 26th, 2026.

On today's show, memory prices may stay high for years as AI demand reshapes the chip market. IBM unveils the world's first sub-1 nanometer chip technology. Google wants AI to use your computer. The bigger question is, should it?

And the fight over AI data centers just keeps on heating up. I'm your host, Jim Love. Those are the headlines. Let's get into it.

One of the world's largest memory chip manufacturers is signaling that memory prices may remain at historically high levels for years. Micron says it signed long-term supply agreements that could lock in today's elevated prices for as much as five years. What's remarkable is why. This isn't being driven by higher manufacturing costs.

It's not because silicon has become more expensive, raw materials are harder to find, factories cost more to operate, or global supply chains remain disrupted. No, it's being driven almost entirely by demand. AI servers require enormous amounts of high-performance memory, and demand has become so strong that manufacturers can focus on producing their highest-margin products instead of competing for lower-margin consumer and business markets. As long as customers are willing to pay premium prices, there is no economic incentive to reduce them.

And that has made memory manufacturers exceptionally profitable. Earlier this week, we reported that employees at South Korea's largest memory companies are already demanding a larger share of those profits through bonuses and compensation. And it's hitting us at the consumer level. Apple has just announced major increases in its prices due to the cost of memory.

But history also suggests another consequence. High margins attract competition. Chinese memory manufacturers, which struggled when prices were lower, now have a powerful incentive to expand production and target lower-cost consumer and enterprise markets. We've already seen something similar happen in AI chips, where export restrictions created an opening for Chinese competitors, and NVIDIA lost that entire market.

For decades the computer industry relied on technology making hardware cheaper every year AI is creating a different kind of market one where demand rather than manufacturing costs is determining price That's great news for today's memory manufacturers. Whether it remains that way once new competitors arrive is a totally different question. IBM says it's developed what it calls the world's first sub-1 nanometer chip technology, introducing a new three-dimensional transistor design called nanostack.

You can't physically imagine how small these circuits are, but the chip packs nearly 100 billion transistors onto a piece of silicon about the size of your fingernail, roughly twice the density of the company's previous 2-nanometer research design. IBM says the technology could eventually deliver up to 50% more computing performance or reduce power consumption by as much as 70%, improvements that could make future AI data centers dramatically more efficient. This is a research breakthrough, not a product announcement.

IBM expects it will take about five years before the technology is ready for commercial manufacturing through partners such as Samsung or Rapidus, assuming development proceeds as planned. What's striking, though, is the timing. Just yesterday, we reported on IBM's $10 billion investment and an ambitious roadmap for quantum computing. Now, only a day later, IBM is making another play for technology leadership, this time in semiconductor design.

Taken together, the announcements suggest IBM is working to reclaim the role it once held as one of the industry's premier research organizations. Rather than competing directly with companies like NVIDIA or TSMC, IBM appears to be positioning itself as the company that invents the next generation of computing technologies, whether those advances happen in quantum computers or the chips that power tomorrow's AI systems. Google this week announced that its latest Gemini platform is moving further to what it calls computer use.

AI that can see what's on a screen, click buttons, type into applications, and complete tasks on behalf of a user. Google says the goal is to let developers build AI agents that can interact with software much the same way a person would. Now, if that sounds familiar, it should. Anthropic demonstrated similar capabilities with Clode's computer use in late 2024, and OpenAI has since introduced operator and other agent capabilities.

The race is no longer about building a chatbot. It's about building digital workers and digital agents. The potential uses are obvious. Imagine telling an AI to prepare your weekly sales report.

It then opens Salesforce, downloads the latest data, updates an Excel workbook, and emails the finished report. Or if you ask it to book a business trip, it compares flights, reserves a hotel, and prepares your itinerary before asking you to approve the purchase. Those examples are attractive because they eliminate repetitive work. But they also raise a much larger question.

An AI can't act on your behalf unless it can authenticate itself somewhere. And that means one of three things. It uses an existing session that you're already logged into. It stores credentials so it can log in by itself later Or it given its own digital identity with specific permissions Each approach introduces new risks If credentials are stored how are they protected Large language models have already shown they can be manipulated through prompt injection and other attacks relatively easily.

So could an attacker trick an AI into exposing credentials or abusing the privileges it's been granted? And even if the credentials remain secure, how much authority should an AI have? Should it be able to approve payments, delete files, or change system configurations? If it makes a mistake, who's accountable?

If it's compromised, how quickly can its access be revoked? Companies like Microsoft already have part of the foundation through technologies like EntraID, role-based access controls, conditional access policies, and service accounts. Those capabilities may prove just as important as advances in AI itself because they provide ways to identify, restrict, and even audit what digital workers are allowed to do. The technology to create AI agents is arriving quickly.

The governance model is still far behind. History suggests that in this case, there will be some hard lessons along the way. There have already been accidental deletions. There will be more of the same.

Just as the early internet taught us the importance of authentication, encryption, and identity management, the next generation of AI agents will almost certainly expose weaknesses that nobody's anticipated yet. So the real challenge isn't teaching an AI to click a mouse. It's deciding who that AI is, what it's allowed to do, and how we make sure it never does more than we intended. And finally, we've been covering opposition to AI data centers for several weeks.

I always struggle about seeing the number of stories. I don't want to bore you with the same story over and over again, but I do want to give you some perspective that what's changing isn't the arguments. It's the intensity. Not just the number of fights that are out there, but the fight over AI data centers keeps heating up to new levels.

One recent example comes from Claremore, Oklahoma, where newly released body camera footage has renewed attention on the arrest of local farmer Darren Blanchard during a public meeting over the Beale Infrastructure's proposed Project Mustang data center. Blanchard supposedly talked too long and was handcuffed and taken to jail after refusing to surrender the microphone when his allotted speaking time expired. Now, if you got thrown in jail for talking too much, I'd still be doing a life sentence.

And normally in these cases, don't the guards just take you and escort you to the door? Speaking afterward, though, Blanchard told 404 Media, I went to a public meeting to speak about an issue affecting my community. I ended up in handcuffs, jailed, and later seeing that moment played and replayed nonstop on television and social media. That is not something you simply move past.

He also warned of what he called a chilling effect, saying, if someone can be arrested after speaking in a public meeting, others may decide it's safer to stay quiet. The incident has become a rallying point for opponents who argue communities are being pushed aside as governments race to approve AI infrastructure But the political consequences are becoming just as apparent In Utah State Senate President J Stuart Adams one of the state's most powerful Republicans, lost his primary election after supporting a major AI data center development near the Great Salt Lake.

At the local level, Box Elder County Commissioner Lee Perry, also lost after voting to advance the same project. Reflecting on the result, Perry told Newsweek, do I think that the data center vote cost me the election? Yes, I do. The defeat suggests opposition to data centers is no longer confined to planning meetings.

It's becoming an issue capable of deciding elections. And developers are responding as well. One of the world's largest proposed AI data center developments in West Jordan, Utah, backed in part by Shark Tank investor Kevin O'Leary, has reportedly been scaled back to roughly half its original size following months of community opposition over water consumption, electricity demand, and quality of life. And now, the debate has even reached popular culture.

In an act of ultimate irony, one of the leaders in this is Polaroid. You may remember the company that made instant photography famous long before digital cameras. and they are still around and they've joined the protest. The company launched an advertising campaign criticizing the water consumed by AI data centers.

Its message is simple, but builds on its own iconic and somehow enduring presence, saying simply existing is an act of rebellion. So what began as local disputes over land, water, and electricity is becoming something much larger. Communities are organizing, elections are being influenced, billion-dollar projects are being redesigned, and even iconic consumer brands are taking sides. One of the key battles in managing a data center is keeping the temperature down so that the place doesn't self-destruct.

It appears that managing the debate about data centers faces a parallel challenge. And that's our show. Heads up with the July 1st holiday in Canada and the July 4th holiday in the U.S.

This is the perfect time for me to take a week of vacation. There will be a Project Synapse on Saturday, but I'll be off next week. although if you're in Minden, Ontario, in that area, you can see my band performing in the Great Canadian Singalong on July 1st. And once again, my thanks to everyone who has come forward to be an early reviewer of my new book.

I've been just thrilled with the excitement and the comments from many of you who've read the original book and enjoyed it. If you've written me, I will have sent you an initial email by the time this show comes out. If you didn't get that, get in touch with me again at TheRealJimLove, TheRealJimLove at gmail.com or at technewsday.

com on the contact us form or reach me on LinkedIn if you're hooked up there. It all works. And if you haven't signed up, what's keeping you? Seriously, you can still join the group of reviewers for my new book, The Compassion Virus.

and join us for Project Synapse on Saturday. Otherwise, I'll be back on Monday, July the 6th. I'm your host, Jim Love. Have a fabulous Friday.