AI Safety Talk and Doom Headlines
AI safety fears have taken over financial media this week. The warning is that people at Anthropic and OpenAI are truly worried about the survival of humanity over the next couple of years. Executives from the major AI companies had to respond. This split them into two camps. One camp said the pace of development should slow down. The other camp - Jensen Huang, President Trump, and Mark Zuckerberg - said there is no real need to slow down.
This same argument has been made before, so it is unclear why this round matters more. The technology world is full of bad predictions. Steve Ballmer said when the iPhone launched that it had no chance of gaining meaningful market share, which proved wrong. The "godfather of AI" predicted in 2016 that radiologists would be obsolete within 10 years, which has not happened. No one wants companies to risk humanity or a Terminator situation, but AI development will keep going.
Macro Backdrop
Macro conditions matter a lot when a sector carries very high valuations. If the 30-year yield keeps climbing, it pressures those valuations. Oil at $100 and above is bad for the macro picture, full stop. Crude in October last traded at 101.80. Two wars are running in key energy markets worldwide, and the Fed just met.
Hiking rates one, two, three, or four times does little when a war disrupts energy supply. Oil needs to come back down and the war needs to stop. If both happened now, financial markets would likely be 10 to 15% higher.
Meta (META)
Meta's (META) big news was its Muse platform. The stock has climbed against the trend while the broader market sold off over recent days. Muse is an AI agent, a step past older chatbots that only answer when asked. It runs on its own on a dedicated cloud computer, can log into your accounts, and carries out tasks for you - booking a tennis lesson, negotiating a lower cable bill or cell phone bill.
Meta's (META) edge is its user base of over 3.5 billion active users, giving it huge data on those users. If Muse gains traction, it should spread fast. Mark Zuckerberg said in a recent interview that Muse will help make and save users money, which would drive quick adoption. The open question is trust: whether people will hand Meta (META) their banking and credit card details and let it act on its own. If early adopters show that trust is there, Muse becomes a big winner for Meta (META).
Shopify (SHOP)
Shopify (SHOP) sits in a debate over whether it gets cut out of the loop or benefits from AI. I lean toward benefit, at least in the short term, because it has built the rails that let AI agents shop efficiently. It built a large catalog that all its merchants connect into, with data structured specifically for AI so agents understand what each merchant sells, and the checkout is seamless.
AI needs to shop in an efficient way. It will not browse the entire internet, since that takes too long. Shopify (SHOP) lets AI agents connect and process purchases quickly. This rail is already profitable and growing. AI shopping has grown triple digits over the past year, which shows merchants are seeing strong demand come through their AI channels. The main risk is oil at $100 or above, which would hurt consumer spending and hit Shopify's (SHOP) business.
Palantir (PLTR)
Palantir (PLTR) helps companies put AI to work effectively, and the results back it up. Companies do not want to hand the keys to AI over to the big players. They want to control their own data and logic. This is driving a shift to "sovereign AI," where businesses balance open-source smaller models with frontier models and avoid handing all their intellectual property to single tech giants.
A real example is Hexion, a specialty chemical maker. It used Palantir (PLTR) to link all its systems - procurement, production, and consumer demand - into one platform, which let it see how one problem could ripple through the whole system. It then added AI agents to watch inventory around the clock and automatically speed up orders during demand spikes. That produced a $300 million EBITDA improvement over three years. Companies see numbers like that, sign up with Palantir (PLTR), and spend more year over year. Things are going well at Palantir (PLTR).
