Rates Hit 5% Again, But AI Held - Now Wall Street Wants Cash Flow, Not Just GPUs

1. The biggest shift this week - AI earnings absorbed a rate shock

At the September 16 FOMC meeting, the Federal Reserve raised its policy rate by 25 basis points to a range of 3.75% to 4.00%. The vote was 12 to 0. The Fed's September median projections point to 2.3% real GDP growth in 2026, a 4.1% unemployment rate, 3.7% PCE inflation, and a year-end policy rate of 4.1%. Since the midpoint of the current range is 3.875%, the Fed's own baseline still leaves room for another hike before year-end.

Inflation is not fully comfortable either. August CPI rose 0.4% month over month and 3.4% year over year, while core CPI, which strips out food and energy, rose 0.3% and 2.4% respectively. Nonfarm payrolls added 162,000 jobs that same month, with unemployment at 4.1%. This is not the backdrop of a recession-driven rate cut.

• 3.75% to 4.00% - the fed funds target range after the September FOMC meeting
• 5%+ - where the US 10-year Treasury yield sits again this week
• $103.87 - Brent crude's close on September 18, still above $100 a barrel

And yet on September 18, the S&P 500 rose 0.17% and the Nasdaq rose 0.40%. Tech stocks propped up the indexes even with the 10-year above 5% and Brent at $103.87 a barrel. That moment matters. The market is not ignoring rates. It looks closer to a judgment that AI-driven earnings growth is still strong enough to offset a higher discount rate.

2. The AI bottleneck is shifting from compute to movement and power management

On September 17, Marvell and GlobalFoundries announced a multi-year agreement to expand production capacity for SiGe semiconductors used in high-speed optical connections inside AI data centers. The deal covers pluggable optical transceivers as well as Near-Packaged Optics and Co-Packaged Optics.

Once GPU counts run into the hundreds of thousands, the speed of a single chip no longer explains system performance. How fast data moves between GPUs, and how little power that movement costs, becomes the bottleneck. That is why the center of gravity in the AI supply chain is stretching from GPUs to HBM to servers to power to networking and optics.

GPU - HBM - Networking - Optics - Power - Cooling - Energy Management

A similar shift is starting in power. The AI Energy Management Alliance, launched with Google, Nvidia and Emerald AI among its backers, is pushing data centers to adjust their power draw in response to grid conditions. Until now the answer to power constraints was "build more plants." The industry is now moving toward treating the data center itself as a giant, adjustable source of demand.

That direction lines up with Vertiv's acquisition of the UtilityInnovation Group. Vertiv is paying $1.45 billion in cash plus up to $1.15 billion in contingent payments to gain microgrid control, on-site generation, and switchgear capabilities. It is a signal that data center power architecture is moving away from pure grid dependence toward a hybrid mix of grid, on-site generation, batteries, and microgrids.

Goldman Sachs Research expects global data center power demand to rise roughly 170% by 2030 from 2025 levels, with grid interconnection waits in some parts of the US stretching up to seven years. GPU supply can scale if more fabs come online, but transmission lines and generation capacity do not expand at the same pace. Power remains the least elastic supply constraint in the long AI cycle.

3. Why Nvidia bought Hugging Face - a chipmaker moves down into the developer platform

Nvidia agreed on September 3 to acquire Hugging Face for $12.9303 billion. Hugging Face is an open AI platform used by more than 18 million developers and researchers.

The point of the deal is not a few extra percentage points of revenue. Nvidia's strategy keeps expanding outward from GPUs into CUDA, networking, AI Enterprise software, models, and now developer platforms. If Nvidia sits at the starting point where developers find, evaluate and deploy models, it can build a lock-in effect stronger than hardware market share alone.

There is a paradox here too. The more layers of the AI stack Nvidia controls, the stronger the incentive for customers to look for alternative suppliers and build their own ASICs. Nvidia's moat is widening, but so may customers' motivation to reduce their dependence on it. Long-term investors need to weigh both forces at once.

4. This week's most important warning - capex can keep rising while stocks rise less

Goldman Sachs' numbers are still strong. Consensus capex among the large hyperscalers is $754 billion for 2026 and $905 billion for 2027, an 83% increase in 2026 over the prior year. Goldman expects these capex-linked companies to account for roughly half of S&P 500 earnings growth this year.

But the latest debate has shifted a notch. Even if capex keeps rising, its growth rate can slow, and as chip supply catches up and prices and margins normalize, the force AI adds to index-level earnings growth can weaken.

This is not a call that AI is collapsing. It is the natural second-derivative problem that shows up as any industry scales. Growth from 700 to 900 is powerful. But growth from 900 to 950, even at a larger absolute size, is a lower growth rate. Equity markets typically react to the rate of change before they react to the absolute level.

5. Market breadth has narrowed - the index is strong, the internals less so

Another signal this week is fund flows. In the week through September 18, US equity funds saw $31.44 billion in net outflows, the fourth straight week of outflows. Large-cap funds alone lost $28.71 billion. Sector funds, by contrast, took in $2.29 billion, and small-caps drew $568 million.

Reading this purely as investors fleeing US stocks misses something. It looks more like money leaving broad index bets and compressing into areas where earnings are visible. The higher rates go, the more the market demands demonstrated profits over a good story.

That is why the industry peak and the stock-price peak need to be treated separately right now. The evidence for AI demand is still strong. But valuations do not need to keep expanding at the same pace. Continued industry growth and wider stock-price volatility can happen at the same time.

6. Robots and space - structural growth signals are strong, but proof still costs a lot

Lucid and Bolt announced a partnership to deploy at least 25,000 autonomous Lucid vehicles across major European cities. The vehicles will use Nvidia's Hyperion technology, and Bolt is targeting a 100,000-vehicle autonomous fleet by 2035. Moving from pilots of dozens or hundreds of robotaxis to supply contracts for tens of thousands of vehicles is a clear shift.

Tesla's Cybercab, by contrast, still faces regulatory scrutiny. NHTSA has asked Tesla to submit documentation by September 30 on the Cybercab's self-certification approach. From here, the numbers to watch for Tesla Robotaxi are not FSD version names but production volume, regulatory approval, actual fleet size, paid miles driven, and revenue per vehicle.

For SpaceX, what happens after its IPO matters more than the listing itself. Following its June IPO, SpaceX secured its first European customer for its Starfall business, which recovers materials manufactured or tested in orbit and returns them to Earth. A 2028 mission is planned to recover up to one metric ton of cargo using Starship and Starfall. It is an early signal that the space industry is expanding beyond launch vehicles and satellite communications into orbital manufacturing and logistics.

Insight Times Editorial Desk