Fed keeps hikes on the table – Amazon, Sandisk and Intel earnings next

Google’s cloud business doubled in size, yet Alphabet shares tumbled after earnings. Microsoft, which also spends heavily on AI, enjoyed a very different reception.

It is a useful warning for tonight’s earnings from SanDisk, Amazon and Intel: investors are still willing to fund the AI boom, but they want to see what the money is earning.

There is another contradiction. The Federal Reserve has paused its rate hikes while oil remains above $110 and diesel is still expensive. The market expects another hike, just not the aggressive sequence it feared a month ago. Meanwhile, refining stocks are falling despite the fuel shortage that helped lift them in the first place.

None of these stories is as straightforward as the headline suggests.

A Fed pause is not a promise of lower rates

The FOMC held its policy range at 3.75–4.00% on 28 October, voting 10–2.

Both dissenters wanted a quarter-point increase. The committee also upgraded its description of economic activity from solid to strong, citing resilient spending, productivity and investment.

At the press conference, Chair Kevin Warsh made a point that goes straight to AI valuations: “Interest rates get in the cracks.”

He was answering whether the enthusiasm for new technology could make its investors immune to higher borrowing costs. His answer was no. Some credit spreads have widened, and markets have become more selective about which investments they finance.

Imagine two companies each asking for another $20 billion to expand AI capacity.

One has customers paying for it already. The other promises that demand will arrive eventually. When borrowing costs rise, investors are less inclined to treat those pitches as equivalent.

Warsh also said investment was broadening beyond AI and that most colleagues did not see the economy as running too hot. That leaves room for the Fed to pause and collect data, without declaring victory over inflation.

50bps hike expectations are being pushed back

The latest CME FedWatch snapshots show the change in sentiment.

December’s probability of reaching 4.25–4.50%, equivalent to 50 basis points of cumulative hikes, has fallen from 42.2% a month ago to zero.

By January, the probability of at least 50 basis points of tightening is 29.5%, down from 69.5%.

For March, it has fallen from about 89% to 67.8%.

So the market is pushing larger hikes further out, not writing off rate hikes.

There is still an 85.6% probability of at least one hike by December, in the 29 October FedWatch snapshot. That could suggest that risk appetite is still staying alive, by a thread, because the Fed is not sounding overly hawkish as expected.

Warsh offered another revealing observation: oil and market interest rates had been rising together for much of the year, but lately bond yields have eased while oil and refinery margins remain elevated.

That can happen when investors start worrying that higher energy bills will hurt growth as well as inflation. It is a complication for the Fed, not evidence that the energy shock has ended.

Diesel is still expensive – So why are refinery shares falling?

Brent was around $115–116 a barrel on the 29 October daily chart, yet the VanEck Oil Refiners ETF (CRAK) has broken beneath its daily 20-EMA band.

The ETF is testing roughly 61.06–62.18 support, with the 20D-EMA (1sd) near $64.81 as the first obstacle to a recovery.

This at least in the short term shows that diesel costs are slinging lower, but now the question is whether the 20D-EMA remains as a source of resistance, or will CRAK begin to rally from support again?

Meanwhile, Brent crude and WTI crude remain elevated, at 115 and 100 levels due to a physical supply shortage.

EIA data for the week ended 23 October show US distillate stocks falling by about 1.5 million barrels to 102.8 million, roughly 10% below their seasonal norm.

Gasoline inventories, meanwhile, rose about 3.9 million barrels.

In plain terms, petrol supplies are recovering faster than diesel supplies. A truck operator does not receive much relief just because refinery shares are down or gasoline margins have narrowed. Diesel still has to be bought, and expensive diesel works its way into freight, farming and construction costs.

This is why an easing of some refinery margins may not give the Fed the inflation relief it wants. It is also why the oil story remains vulnerable to another geopolitical disruption, even if diplomatic headlines briefly push crude lower.

SanDisk: Are these profits real, or just the best part of the memory cycle?

SanDisk sells the NAND flash storage used in data centres, enterprise SSDs and devices that support AI workloads. It is a different business from manufacturing the high-bandwidth memory attached to AI training GPUs, but increased demand for storage and inference can still be lucrative.

The numbers from its August earnings release were extraordinary: quarterly revenue of $8.97 billion, up 51% from the previous quarter, and an 84.6% gross margin. But management said two-thirds of the revenue increase came from higher selling prices. Only one-third came from volume.

That is where a sceptical investor would interrupt. If customers are paying twice as much for memory during a shortage, how much of that profit survives when supply catches up?

SanDisk’s answer is to sign longer-term customer agreements with volume commitments and pricing protections. The promise is a less volatile business than the old memory cycle.

The catch is that investors cannot assume today’s extraordinary margins are permanent until those contracts have been tested through weaker prices.

The cash numbers need a second look too. SanDisk generated $7.08 billion of conventional quarterly free cash flow, against just $43 million of direct property-equipment purchases.

But its own adjusted FCF was $5.04 billion, after accounting for customer prepayments and Flash Ventures activity. It also uses a manufacturing joint-venture structure with Kioxia, meaning the direct capex figure does not describe every investment needed to maintain supply.

The company had around $4.76 billion in cash and no reported debt at its last fiscal year-end.

So the immediate question is not whether it can keep the lights on. It is whether cash flow remains this strong when customers stop paying unusually high prices or advancing cash.

On the daily chart, a possible double bottom and bullish Stochastic RSI divergence are developing around $1,278. But price still needs to reclaim the $1,432 area, then the upper 20-day EMA band near $1,600.

Tonight’s fiscal first-quarter report needs to show how much growth comes from actual shipments, where gross margins are headed and whether contracted demand remains strong.

Amazon: AWS is booming – Why is free cash flow negative?

Amazon’s AI exposure is much more direct. AWS sells computing capacity; Trainium provides its own AI chips; Bedrock offers access to models from companies including Anthropic and OpenAI. Amazon is both a supplier of AI infrastructure and one of its biggest investors.

Its second-quarter results were strong on the operating side. AWS sales rose 37% to $42.2 billion, producing $16.6 billion in operating profit. Across Amazon, operating profit increased 43% to $27.5 billion.

Then comes the question investors cannot ignore: if business is this profitable, why is free cash flow negative?

Amazon generated $45.4 billion of operating cash flow in the quarter, but spent $54.2 billion buying property and equipment. Over the trailing 12 months, operating cash flow reached $161.4 billion while net equipment spending reached $169.0 billion. That left reported trailing free cash flow at negative $7.6 billion.

This does not mean AWS is losing money. It means Amazon is buying data centres, chips and other assets faster than its businesses currently generate cash. The company also ended June with roughly $123 billion in cash and marketable securities, against $128.9 billion in long-term debt, up sharply from $65.6 billion at the end of 2025.

For investors, the real question is what return Amazon will earn on the next wave of capacity. An AWS growth beat would be welcome, but margins, 2027 spending plans and a credible route back to positive free cash flow could matter more.

There is an earnings-quality wrinkle. Amazon’s previous GAAP net income included a $53.4 billion non-operating pre-tax gain, primarily from its Anthropic investment. That is not the same as earning $53.4 billion selling cloud services. Operating profit and cash flow provide a clearer starting point.

The weekly chart remains within a rising channel. After an approximately 84-day pullback, the 50-week EMA around $243 is an important support reference. A hold would preserve the broader uptrend; a break would leave investors more exposed to disappointing guidance. The oscillator is not yet deeply oversold, so support alone is not a reason to assume the selling is finished.

Intel: The turnaround is improving – Can the factories pay for themselves?

Intel is exposed to AI through Xeon server CPUs, custom silicon and advanced packaging, alongside its attempt to win outside chipmaking customers for Intel Foundry. Unlike SanDisk, its biggest uncertainty is not simply the price of what it sells. It is the cost of making it.

Second-quarter revenue rose 25% to $16.1 billion, and Data Centre and AI sales increased 59%. Better yields helped the non-GAAP gross margin reach 41.8%. Those are genuine signs of improvement.

But Intel Foundry still recorded an operating loss of about $2.09 billion. Most of the division’s reported revenue came from manufacturing for Intel itself, rather than independent customers. A foundry cannot become the next great profit engine solely by invoicing another part of the same company.

Its cash flow also needs careful reading. Q2 operating cash flow was $7.01 billion and gross capex was $2.65 billion, which gives conventional FCF of about positive $4.35 billion.

Yet Intel’s adjusted FCF was negative $8.42 billion after significant partner cash outflows linked to restructuring ownership of a manufacturing venture and finance-lease payments.

That is not evidence that ordinary chip operations burnt $8.42 billion in the quarter. It is evidence that restructuring and financing advanced manufacturing can require vast amounts of real cash, even when operating results improve.

Intel had about $29.7 billion in cash and short-term investments against $50.5 billion in debt at June-end, before subsequent equity fundraising.

The coming results must answer two questions: are factory yields improving quickly enough to lift margins, and is outside demand for future manufacturing processes becoming firm customer business?

Without that, a recovery in server-processor sales may still be carrying a costly foundry operation.

The daily chart is recovering from oversold territory, but price remains below the 20-day EMA band. The old June support and neckline around 98.33–106.48 is the next major test. 

$79.62 remains an important lower support reference. Better earnings would strengthen the rebound argument; unconvincing foundry economics would make resistance harder to clear.

So what actually changes the investment case?

The previous day’s results provided a warning. Google Cloud revenue rose 103% to $30.7 billion, but Alphabet reported negative $9.1 billion of quarterly FCF and said its AI capex would increase further next year. Microsoft’s revenue and operating profit grew while Azure sales rose 48%. The market drew a sharp distinction between AI growth and the cash available after paying for it.

That is the thread connecting the Fed, diesel and tonight’s earnings. Expensive energy pushes up the cost of building and operating infrastructure. The Fed’s slower expected rate path gives investment plans some breathing room, but borrowing is not free. And investors are beginning to demand proof that AI spending can eventually repay itself.

For SanDisk, watch volume growth, pricing and adjusted cash flow.

For Amazon, watch AWS margin and what management says about capex.

For Intel, watch Foundry losses, external customers and manufacturing yields.

Those are the numbers that could justify a stronger earnings reaction, or expose the limits of the AI trade.

The Nasdaq 100 remains above its 100-day EMA band (0.5 standard deviation) on the 29 October daily chart, with a potential bullish momentum divergence.

That support keeps a recovery possible for now. A decisive break below it would say the market is no longer prepared to give expensive growth stories the same benefit of the doubt.

The verdict is conditional. If the Nasdaq holds its band and tonight’s reports show that growth is turning into repeatable cash generation, there is room for the recovery to continue.

If the numbers rely on temporary pricing, non-cash gains or yet another increase in capital commitments, a better earnings headline may not be enough. The market has already shown that it is prepared to punish an expensive AI story even when demand is booming.