By Nicholas Hamner
Investment Advisor Representative
[email protected]
AI is in the news again—as it seemingly always is. The past few days saw several developments worth noting, both for what they signal today and what they could mean for the future.
The big one is a shift in investor outlook on Big Tech’s AI spending. We started to see a divide between companies that manufacture the hardware used in AI data centers—Nvidia, Broadcom, and Micron—and the companies spending hundreds of billions of dollars for those data centers: Alphabet, Meta, Microsoft, and Amazon, who are now called “hyperscalers.”
For most of our recent memory, both groups grew together. Investors could buy the companies making AI hardware and the companies building the data centers, confident that growing demand would benefit both. Now, however, the hyperscalers have collectively pledged nearly a trillion dollars toward AI investments and their investors are asking a more difficult question: How much is too much to spend when we haven’t seen a profit yet?
Last week, the major hyperscalers saw market declines. Related businesses have also felt pressure, like Caterpillar, whose equipment helps build data centers. With the builders and buyers feeling more market pressure, it raises an important question for the hardware manufacturers: What happens when their customers’ wallets don’t open so frequently?
That’s not to say that these companies can’t or won’t ultimately benefit from their investments. It does indicate, however, that a more selective and cautious investment approach may be necessary.
Another development worth noting: AI models are escaping their safeguards. Last week, OpenAI announced that one of its experimental models found a way out of its test environment, accessed the open internet, and hacked another website to more easily accomplish its given objective. The immediate impact was limited to a single website, but the implications are much broader.
As security expert and tech CEO Marius Hobbhahn put it, “If a model of this capability level cannot be contained, what should we expect for future, much more powerful models?” Reports that this may not have been the first instance of a model escaping its sandbox add to those concerns.
An MIT study, published July 22, 2026, surveyed 272 international AI experts across 24 risk categories. The experts looked at both the potential harms and the likelihood of those harms occurring. Under a scenario in which governments and technology companies cooperate to strengthen safeguards, five categories—including a rogue AI developing cyber and physical weapons—were assigned at least a 10% chance of causing one million deaths or $100 billion in financial damage by 2030.
When those same experts were asked to consider a future in which business continued as it is today, the number grew to 18 of the 24 categories.
In some ways, AI today feels like nuclear energy must have felt in the 1940s and 1950s. The potential is enormous. Even today, nuclear power remains one of the most efficient and environmentally friendly energy sources available when properly managed, but it also presents serious risks and difficult questions. AI carries a similar challenge: extraordinary promise alongside potentially drastic consequences.
International competition shows no sign of slowing, meaning both rapid advancement and meaningful restraint carry uncertainty. A measured, informed approach is critical. Decisions driven by emotion—whether fear of missing out or fear that risks are spiraling out of control—can be costly.
That is why it is important to work with a trusted advisor focused on your best interests. All of us at Franklin Retirement Solutions—Peter, Rob, Kyle, Jeremy, Austin, and myself—are watching developments in technology and AI from every foreseeable angle. No one wants to be caught lagging behind, but no one wants to be overexposed either. If you want to talk through what these developments could mean for your financial plan, just let us know.