The market spent the week celebrating record highs on Tuesday, only to spend the rest of the week realizing who has to pay for them. The S&P 500 slipped 0.47% and the Nasdaq dropped 1.25% on Thursday before clawing back ground on Friday. For the full week, the S&P 500 gained about 1.2% and the Nasdaq added about 0.6%, while small caps in the Russell 2000 dropped roughly 0.9%. That divergence tells the real story. The artificial intelligence trade has stopped being a simple argument about computing demand. It has turned into an uncomfortable conversation about the cost of money.
You can see the shift across the bond market. The 30-year Treasury yield touched 5.71%, its highest level since 2002, while the 10-year yield finished near 5.24%. That is the borrowing benchmark that ripples directly into your mortgage rate, your car loan, and corporate credit lines. Brent crude pushed back above $100 a barrel amid Middle East tensions, while the Federal Reserve sits at a 3.75% to 4.00% benchmark rate following a September hike under Chair Kevin Warsh. Meanwhile, the real economy showed genuine fatigue. The September payroll print revealed just 29,000 jobs added against the 84,000 expected, with unemployment ticking up to 4.2%.
Yet tech infrastructure bills keep expanding. Samsung posted a ninefold profit surge and Taiwan Semiconductor Manufacturing Company grew monthly sales by 51%, but neither stock rallied. Amazon Web Services raised reserved GPU server prices by roughly 15% on Wednesday, marking its third price hike this year. When the price of electricity, silicon, and long-term borrowing goes up simultaneously, corporate balance sheets start feeling the pinch.
The four names on this week's list represent the four corners of that invoice. Alphabet pays the bill. Micron charges for the hardware. Vistra meters the electricity. RTX sells to a military customer that ignores interest rates entirely.
Before diving into this week's picks, we check our receipts. Seven days ago, our four focus stocks were waiting to replace market debates with concrete financial numbers. Exactly none of the four delivered their scheduled quarterly figures this week, but one delivered a massive strategic shock. All four major earnings verdicts still lie ahead within the next month.
Constellation Energy produced the week's biggest fireworks without reporting earnings. On Tuesday, Google signed a landmark 20-year pact for roughly 890 megawatts of new nuclear capacity through reactor uprates, alongside a 15-year deal for another 2,700 megawatts. Constellation committed over $4.3 billion to fleet upgrades on the news. The stock jumped roughly 15.8% for the week, leading our group, though it remains down about 15% year to date. Its test remains proving that full-year adjusted earnings track within its $11.50 to $12.50 per share guidance.
Amazon added about 4.3% over the week and remains near the top of its range, up roughly 14% this year. The company still needs to prove AWS revenue can grow at or above 35% with margins in the high 30s. SoFi barely moved, rising 0.2% over the week while lingering roughly 40% lower year to date. Its date with destiny arrives on October 27, where it must show earnings per share heading toward its $0.60 full-year target. Finally, Netflix climbed about 4.8% off its 52-week low, leaving it down roughly 25% on the year. Its test arrives on October 20, where operating margins must hold at or above 31.5%.
1. Alphabet (GOOGL) — The Full Stack
Alphabet is the parent company of Google. It collects vast revenues by placing digital advertisements across Search and YouTube, while renting computing infrastructure through Google Cloud. The clearest way to understand the business is that Alphabet owns the entire digital skyscraper. Rivals rent land, buy specialized chips from third parties, and license outside models. Alphabet designs its own custom processors, builds the physical data centers, trains the underlying Gemini models, and owns the consumer apps where billions of people interact daily.
That complete ownership generates extraordinary scale. For the June quarter, revenue rose 24% to $119.8 billion, marking twelve consecutive quarters of double-digit expansion. Core advertising and subscriptions brought in $94.5 billion, with Search climbing 17% and YouTube ad sales rising 13%. Google Cloud surged 82% to $24.8 billion while doubling its operating margin to 35.6%. Future signed cloud contracts hit $514 billion. The Gemini mobile application counts 950 million active monthly users, and developers push 22 billion tokens through its programming interfaces every minute.
On paper, diluted earnings jumped to $9.11 a share. But strip out a $99 billion non-cash accounting gain on equity holdings, and real earnings grew an impressive 23% to roughly $2.85 a share.
Despite those figures, the stock trades roughly 14% below its peak and sits around 23 times forward earnings. That valuation is lower than Microsoft's 26 times multiple. The market's lingering skepticism comes down to capital expenditures. Two years ago, quarterly capital spending was $13.1 billion. In June, it hit $44.9 billion. Management raised full-year spending estimates to between $195 billion and $205 billion, and Wall Street projects $257 billion for next year.
Free cash flow—the cash left over after paying all operating costs and capital investments—swung to a negative $5.9 billion in the second quarter. Investors are debating whether Alphabet is cementing an unassailable technological lead or simply burning through cash.
The catch is that Google is leaning on outside capital to finance this buildout. The company issued $49.6 billion in equity and $20.3 billion in notes during the June quarter while pausing its share repurchase program. Off-balance-sheet commitments include roughly $43.8 billion backing third-party data center developments and $7.6 billion in supplier guarantees. Legal battles add more friction. A federal judge recently allowed a $3.2 billion publisher ad-tech antitrust damages claim to move forward, which could triple under antitrust statutes. If massive infrastructure outlays fail to create software profits that outrun advertising depreciation, Alphabet's returns on invested capital will drop.
What we're watching: Third-quarter results scheduled for Wednesday, October 28. Bulls need to see cloud revenue growth maintain a 60% pace with margins above 35%, alongside positive free cash flow. A guide capping next year's spending near consensus would relieve pressure. The kill condition is two consecutive quarters of negative free cash flow paired with Search growth slipping below 10%.
Our take: Accumulate. You are paying an ordinary market multiple for the only tech giant that controls its silicon, models, and distribution. We would reconsider if free cash flow remains negative without a clear ceiling on infrastructure outlays.
2. Micron Technology (MU) — The Memory Toll Booth
Micron Technology designs and manufactures memory chips, specifically dynamic random-access memory, known as DRAM, and flash storage, known as NAND. In the AI ecosystem, high-bandwidth memory, or HBM, sits stacked right alongside processing chips to prevent data bottlenecks. Think of high-end graphics processors as multi-million-dollar sports cars. Micron sells the specialized racing fuel and tires. Without stacks of ultra-fast memory feeding gigabytes of information per second, advanced computing engines cannot function.
The pricing power behind that hardware is stunning. In its fiscal fourth quarter, revenue surged 379% year over year to $54.2 billion. DRAM accounted for $39.8 billion of that sum. This boom is driven by price rather than pure volume. DRAM bit shipments grew in the mid-single digits, yet average selling prices climbed in the high teens from the prior quarter. That dynamic expanded gross margin—the share of revenue retained after production costs—to 87.0%.
Non-GAAP earnings hit $33.42 a share, generating $33.2 billion in free cash flow during a single quarter. Micron holds $73.5 billion in cash against just $5.2 billion in debt. Crucially, 26 long-term customer agreements now lock in over 35% of revenue through 2030, with future contracted performance obligations standing near $150 billion.
Yet the stock trades at roughly six times forward earnings, sitting about 18% below its high despite tripling over the past year. That valuation is an anomaly. The broader market refuses to treat memory as a durable growth business. Historically, memory has been a volatile commodity where chipmakers overbuild factories, cause a supply glut, and watch pricing collapse. During the last major downturn into 2019, DRAM prices plunged roughly 80%. Competitor SK Hynix trades even lower, under four times forward earnings.
The current multiple indicates that Wall Street believes these peak profits will evaporate once industry supply catches up.
That fear represents the core risk. Micron is planning roughly $25 billion of capital spending in the first half of fiscal 2027 alone. Rivals Samsung and SK Hynix are also pouring tens of billions into new fabrication plants. When those cleanrooms open, supply will jump just as the industry transitions to next-generation HBM4 designs.
Because Micron's recent earnings spike came almost entirely from price hikes rather than unit growth, any downward shift in memory pricing would compress margins immediately. Furthermore, inventory holding periods ticked up by nine days last quarter to 129 days. If enterprise hardware spending slows, those components lose value quickly.
What we're watching: Fiscal first-quarter results due in mid-to-late December, following an ex-dividend date on October 14. We want to see revenue reach the guided $61.5 billion mark with gross margins holding near 86.25%. The trade breaks down if sequential memory prices turn negative, inventory days continue climbing, or hyperscale cloud providers trim their capital spending budgets.
Our take: Accumulate. Trading at roughly six times earnings, the stock already prices in a harsh cyclical downturn, ignoring multi-year customer contracts that already cover more than a third of its revenue through 2030. We would exit if contract commitments fail to prevent sequential price drops.
3. Vistra (VST) — The Bill Collector
Vistra is the largest competitive power generation company in the United States. It operates roughly 44 gigawatts of generation capacity across nuclear, natural gas, coal, solar, and utility batteries. It also runs a retail power unit that delivers monthly electric bills to approximately five million homes and businesses. Most investors focus on who builds power plants. Vistra owns the plant, generates the physical electricity, and collects the monthly subscription from the end customer. That integrated retail setup protects the bottom line when wholesale spot energy prices fluctuate.
Recent operational performance shows the strength of that model. Second-quarter ongoing adjusted EBITDA—earnings before interest, taxes, depreciation, and amortization—grew 31% to a record $1.767 billion. The retail business generated the largest share at $773 million. Fleet availability held above 97% during peak summer heatwaves. Nuclear capacity remains the company's prize asset.
Vistra secured 20-year power purchase agreements with Meta for 2,609 megawatts and Amazon for 1,200 megawatts from its Comanche Peak reactors. In the latest PJM capacity auction, where grid operators buy future reliability commitments, Vistra cleared 10,566 megawatts at an average of $333.44 per megawatt-day, bumping right against the market price cap.
Despite these tailwinds, the stock sits roughly 25% below its 52-week peak and is essentially flat for the year. It trades around 15.5 times forward earnings, compared to peer Constellation Energy at roughly 21 times. The discount exists because Vistra aggressively hedged its future power output. The company has locked in prices for 100% of its generation for the rest of this year, roughly 94% for 2027, and about 72% for 2028. Hedging operates like an insurance policy: it guarantees cash flows, but it prevents the company from capturing sudden price spikes caused by surging data center demand.
That structural cap creates a real timing problem. Management recently guided full-year 2027 financial opportunities toward the lower end of its $7.4 billion to $7.8 billion range. Reported quarterly revenue declined 5.5% year over year. The balance sheet carries leverage around 2.6 times debt to operating earnings, supported by a recent $1.5 billion junior note issuance to help fund up to $5.0 billion in planned growth outlays. Furthermore, independent grid monitors recently raised market-power objections regarding Vistra's planned acquisition of Cogentrix. With almost all 2027 power already committed at fixed rates, shareholders must wait until 2028 or 2029 to see the financial upside of recent AI deals.
What we're watching: Third-quarter earnings on Friday, November 6, followed by PJM reliability procurement decisions on December 2. We want to see capacity awards clearing near the expected $400 to $555 per megawatt-day range. The thesis fails if full-year adjusted EBITDA misses the $6.8 billion floor or if regulatory authorities block the Cogentrix deal.
Our take: Wait for a better entry. Vistra is an exceptional operational utility, but heavy hedges mean surging power prices will not reach the bottom line for two years. We would become buyers after a valuation pullback toward the bottom of its historical range.
4. RTX (RTX) — The Bullet Counter
RTX operates as a diversified aerospace and defense conglomerate across three major segments. Collins Aerospace and Pratt & Whitney manufacture aircraft electronics and commercial jet engines. Like selling an expensive printer and profiting from replacement ink cartridges for twenty years, jet engine builders earn steady, high-margin cash on spare parts and overhaul maintenance. Meanwhile, its Raytheon defense arm manufactures guided missiles, radars, and air defense systems. RTX acts as an essential repair shop for global commercial airlines and the primary ammunition depot for Western military rearmament.
Defense spending commitments are filling the order book faster than factories can build. Second-quarter sales rose 14% to $24.7 billion, while free cash flow reached $2.878 billion. Raytheon expanded operating profit by 29% as its book-to-bill ratio—the dollar value of new orders received divided by orders shipped and billed—hit 2.42 for the quarter and 1.77 over the trailing year. Total company backlog reached $289 billion, roughly three times annual sales. In recent weeks, the Pentagon awarded Raytheon a $22.9 billion Tomahawk missile contract to lift output toward 1,000 units yearly. It booked a $20.7 billion AMRAAM deal, plus up to $6.3 billion for Standard Missile-3 interceptors.
Yet RTX trades roughly 18% below its summer high and is up only 2.5% this year, lagging the broader market. The stock commands a forward earnings multiple near 24 times. That valuation represents a noticeable premium over direct defense peers like Lockheed Martin and Northrop Grumman, which trade around 16 times earnings. Investors are unwilling to push the stock higher because massive contract headlines have become routine. When you pay a premium valuation, flawless execution must follow, and the market wants proof that production lines can actually handle this backlog.
The operational bottleneck is severe. A backlog is an accounting record of future promises, not immediate cash in the bank. Roughly 45% of RTX's commitments represent commercial engine maintenance contracts stretching over two decades, with only a quarter converting to revenue within twelve months. To fulfill its missile deals, Raytheon must increase Tomahawk production sixteen-fold from historical rates of 60 per year while doubling AMRAAM output. The defense industrial base faces chronic shortages of rocket motors, titanium castings, and specialized technicians. Lingering legal issues also present headwinds: the Securities and Exchange Commission continues investigating company disclosures tied to its 2023 engine metal defect crisis.
What we're watching: Third-quarter financial results on Tuesday, October 20. We need to see full-year free cash flow tracking toward the top of its guided $8.50 billion to $8.75 billion range, with Raytheon segment margins pushing past 12.6%. The warning sign would be Raytheon's quarterly book-to-bill falling below 1.0 or commercial aftermarket maintenance turning negative.
Our take: Wait for a better entry. Government rearmament budgets are guaranteed, but RTX's 24 times earnings multiple leaves zero room for manufacturing delays. We would wait for the multiple to compress toward peer levels before committing fresh funds.
The bottom line
The conversation driving equity markets has shifted from how fast the artificial intelligence buildout can grow to who carries the financial burden when borrowing costs climb. Alphabet proves that owning every layer of technology requires massive capital outlays that can temporarily strain corporate cash generation. Micron captures the physical hardware bottleneck by charging peak prices for essential memory, even as the market treats those windfall profits as cyclical. Vistra generates and bills for the electricity powering data centers, but comprehensive hedging shields its earnings from immediate price surges. RTX operates on a separate plane altogether, supported by government defense budgets that do not depend on prevailing interest rates.
We will track all four positions against their stated milestones over the coming weeks, grading their execution in plain view.
Analysis, not advice. We may own these names and we may sell them at any time.
