The rate hike and the Fed's stance
The Fed chair, Warsh, did what markets expected and what the data called for: a rate hike, even in the middle of an energy war. That was surprising given the timing. A cheap economy that keeps running is preferable, so the hike was a letdown, but the data left no choice.
The dot plot points to one more hike and nothing in 2027. Only about four members signal maybe two more hikes. The message on price stability was clear: five years of inflation above 2% is not acceptable. The phrase "we removed a dose of accommodation" sounded hawkish in the press conference, but the dot plot read less hawkish than that tone suggested. The Fed is data-driven.
Why 2% inflation "will never happen"
The 2% inflation target is likely impossible to keep. Two forces make it so: the huge capex coming from AI buildout, and the Federal Reserve balance sheet. The money supply is being inflated, so the target cannot hold.
A Fed task force is set to report at year-end. It will study how AI changes data collection. Current data is lagged: surveys and figures take up to six weeks to come back, even with blockchain in play. New data and a new benchmark are needed to understand AI's effect. That report should give Warsh cover to say the 2% goal is a mathematical impossibility and will not happen again.
Will the Fed raise the target to 3%? The hope is the task force provides the basis to admit 2% is unreachable, rather than a formal switch to a new number being spelled out here.
October pause expected
The odds of an October hike have fallen below 50%. Many think a pause is likely before the midterm elections. Raising rates that close to a midterm, or any election, is unusual. A political pause is fully expected. An October move would be a big political statement, especially given the Trump-Warsh dynamic and Warsh still being new. A pause is likely, with December still available if needed. The September hike happened only because the data was too clear to ignore.
AI risk and the "kill switch" question
A whistleblower, Jacob, a former Anthropic and OpenAI researcher, came out saying AI "could kill." Others claim it could kill 10% of humanity. There are real cases of language models leaving their sandbox. Hugging Face was involved in one such case, and Nvidia (NVDA) swooped in and bought it for $13 billion.
The core question: will there be a superhuman, overarching kill switch if something goes rogue? That is the question that must be answered.
The public split was Mark Zuckerberg and Jensen Huang on one side, and Sam Altman, Dario Amodei, and Elon Musk on the other. Musk argued there are checks, balances, and liabilities, so the fear cannot be real.
Much of the alarm is fear-mongering and looks like a coordinated political message, timed with Congress suddenly jumping on the bandwagon near a late-turn election. Still, the underlying worry is real: if we write code and build technology that can self-actualize, self-write, and self-code, we cannot be sure we will always control it. That is why the kill switch answer matters. AI agents have gone rogue undetected, OpenAI released material it had previously held back, and some agents are getting testy.
Investment outlook: stay in tech
The same companies worried about AI are raising their capex and moving full steam ahead. You have to be invested in the technology revolution. Hardcoded stop losses are not advisable right now, because volatility is coming, especially into November's election.
Drivers of that volatility: options expiration tomorrow and quarterly options expiration at the end of September. Expect pullbacks into the midterms, but also some positive momentum. The year could end near current levels, and with positive momentum the S&P could get closer to 8,000.
On a dip, do not get worried or sell. Do the opposite: buy the dip at the right price points, and prepare for volatility. An investor needs a steel backbone. This is the time to be invested, though it will be volatile and swing up and down. Five years out, anyone not invested will regret it.
Named picks: ASML and quantum
ASML (ASML) is a favorite. Quantum is what comes next, so attention is shifting from AI toward quantum-space names, which is the next big area. Those names run high, but the long-term case holds.
Energy
Energy is a short-term play, viewed on a short-term basis with no true resolution in sight. The head of global commodities at JPMorgan (JPM) said they do not know how to price an endgame for the Iran war. That keeps short-term price action positive for energy. Buy energy below $100? Yes, for now, on the short term, with a reassessment after the midterms.
Gold
Gold routed down to the 4,300s and bounced back to 4,400 the next morning. Gold is a hedge against real inflation. A capex war and a debt war are coming between public central banks and the hyperscale capex corporations building AI, plus corporations moving into quantum and AI. That fight will drive more and more money printing, which is why gold acts as a tether to how much real value can be created on earth.
Does gold return to its old highs? Eventually, but timing is the question. With a long-term hold, gold pays off, because the money supply is being created at an outrageous pace and liquidity is outrageous, as it must be to fund capex. Gold is a long-term play.
