The Focus List: Owning the Contract Instead of the Projection

The artificial intelligence trade did not break this week, but it did lose its primary face. Jensen Huang resigned after thirty-three years running Nvidia, handing interim leadership to finance chief Colette Kress while the board began an outside search. Nvidia dropped as much as 14% before closing more than 4% lower that session. Yet Broadcom gained about 5% and AMD climbed roughly 6%. The capital did not pack up and head for cash; it simply walked away from an individual founder.

A second signal landed in the desert. SpaceX committed more than $74 billion to build two massive computing campuses in Texas and Arizona. The deciding factor was not cooling technology or property taxes. It came down to power and local permits, and the two sites solved them in opposite ways. Texas will generate its own electricity on company land. Arizona is negotiating with rural electric cooperatives, and still needs a county vote on December 9.

Then came the macro headwind. A resilient November payroll print prompted futures markets to price in real odds of a Federal Reserve interest rate hike in December. That single shift raises the price tag on the debt funding every data center, turbine, and server rack in the country. The through-line across all three events was unmistakable: investors spent the week paying a premium for cash that arrives via long-term contracts, hard order backlogs, and non-negotiable subscriptions. At the exact same time, they marked down businesses whose next leg of growth depends on an executive's optimistic slide deck.

Scorecard

Our focus list from last week took this exact rotation in stride. None of the four names has reached its formal decision window yet, with key catalysts bunched between December and February, but the tape gave us an early read on who has pricing power.

Broadcom was the standout, leaping about 18% on the week for its sharpest rally in months after locking in December 9 for its fiscal fourth-quarter print. That session will test both its $21.7 billion quarterly AI semiconductor target and its $115 billion target for fiscal 2027, just twenty-four hours before the Fed meets.

Amazon gained about 3% to sit roughly 6% below its peak after leadership reiterated that cloud revenue should compound above 40% through 2028, setting up a pivotal holiday update in early February. SoFi dropped about 5%, pressured alongside the broader fintech space by rate anxieties. It sits nearly half below its high while we await the December rate verdict, its ninety-cent earnings target, and its 4% default line. Zscaler slipped about 3% after reaffirming its first-quarter sales range of $935 million to $939 million for early December, reminding us that waiting for execution beats chasing a software bounce.

1. Constellation Energy (CEG) — The Toll Road

Constellation owns the private highway that artificial intelligence clusters must use to reach the power grid. It has started selling decade-long express passes instead of collecting change at the booth.

The company is America's largest nuclear operator, managing roughly 55 gigawatts of nuclear, natural gas, and geothermal generation after absorbing Calpine. It sells electricity across competitive wholesale markets. The real advantage is selling directly to hyperscalers, the huge technology platforms building global computing networks, through long-term power purchase agreements, or PPAs. Those contracts lock in power deliveries at fixed rates for decades.

Third-quarter numbers proved the financial leverage of that model. Operating revenue climbed roughly 45% to $9.55 billion, while adjusted operating earnings reached $5.06 a share against $3.04 a year earlier, a rise of about two-thirds. Operating earnings guidance for the full year stepped up to between $12.50 and $12.80 a share, marking the second upward revision this year.

More importantly, Constellation signed deals adding more than 1,000 megawatts of clean generation—roughly the output of a brand-new commercial nuclear reactor—without charging everyday ratepayers. Google signed a 15-year commitment for 2,700 megawatts, plus a 20-year pact funding over $4.3 billion in plant upgrades across eleven grid facilities. Amazon inked a 20-year, 690-megawatt contract underpinning $3 billion of capital work at Calvert Cliffs.

The stock gained about 12% on the week, climbing back to its 200-day moving average yet still sitting roughly a quarter below its highs. The market hesitates because Constellation is the most expensive merchant generator on the board. It trades near 21 times forward earnings, meaning the share price divided by expected per-share earnings. That compares with roughly 14 times for Vistra and under 10 times for NRG. It also shoulders about $22.8 billion in net debt, and free cash flow that barely clears a billion dollars leaves its cash multiple perched in the seventies.

The bear case focuses squarely on politics. Constellation's valuation relies entirely on its contract backlog, and those contracts tie directly into regional capacity auctions—the market mechanism where power plants get paid simply to guarantee standby supply. State politicians are already pushing back against capacity auctions as residential electricity bills jump. If regulatory authorities freeze local data-center connections, the pipeline of multi-decade agreements will stall out, leaving equity holders holding Calpine's debt pile on the lower multiple of a standard utility.

What we're watching: Full-year earnings around February 24, 2027, when leadership issues its initial 2027 outlook. The bull thesis requires base earnings expansion of at least 20% and another gigawatt of contracted deals priced comfortably above the federal nuclear tax credit floor. The bear case is confirmed if 2027 profit guidance breaks below the $12.50 to $12.80 run-rate or if the next grid auction clears at price caps without fresh customer deals in hand.

Our take: Accumulate. You are paying up for locked, inflation-protected cash flows that stretch across two decades, just as the wider market is discovering that founder stability and debt financing are no longer guaranteed. We would reverse this view if contract signings freeze for two consecutive quarters or if net debt stops falling.

2. RTX Corporation (RTX) — The Parts Counter

RTX sells the aircraft engine near production cost so it can invoice the airline for replacement hardware, routine inspections, and shop maintenance for the next thirty years.

The enterprise runs three industrial engines. Collins Aerospace produces cockpits and cabin interiors. Pratt & Whitney builds commercial turbofans alongside long-term fleet service agreements, and Raytheon manufactures guided interceptors, surveillance radars, and air defense platforms. The crown jewel is the aftermarket—the predictable stream of spare parts and scheduled maintenance that airlines are legally required to buy as long as planes stay in the air.

The latest quarterly print showed commercial travel offsetting industrial pain. Sales rose 11% to $25.0 billion, assisted by 12% organic revenue expansion. Adjusted per-share earnings hit $2.26, rising 33%, though $0.25 of that boost came from a one-off trade tariff recovery. Total backlog climbed 36% to an unprecedented $340 billion, split between $176 billion commercial and $164 billion defense. That was helped by more than $81 billion in fresh bookings, including production awards for the Tomahawk and AMRAAM missile systems.

Management lifted full-year adjusted sales targets to between $97.0 and $97.5 billion, with free cash flow projected at $9.0 billion to $9.2 billion. S&P rewarded that balance-sheet resilience on November 13 by bumping RTX's debt rating to A- from BBB+.

The equity rose about 5% on the week, sitting roughly 12% under its peak and slightly above its 200-day trendline. Critics argue that RTX is no longer a value stock, trading near 24.5 times forward earnings compared to roughly 18 times for defense peers Lockheed Martin and Northrop Grumman. Investors are paying a premium for commercial aerospace exposure, leaving no safety margin if military assembly lines stall.

The real catch lies in the plumbing of its cash flow. While adjusted earnings jumped 33%, quarterly free cash flow dropped roughly 47% to $2.1 billion. Headline numbers benefited from that one-time tariff clawback, while a $363 million hit tied to an airline bankruptcy was excluded from adjusted totals. Furthermore, Raytheon booked major missile wins as undefinitized contracts—agreements where production begins before final pricing is legally cemented. If factory bottlenecks or supply chain inflation persist, the actual profitability inside that $340 billion backlog will disappoint.

What we're watching: Fourth-quarter results on January 26, 2027. The bull case needs at least $3.1 billion of fourth-quarter free cash flow to land inside the annual $9.0 billion to $9.2 billion range. It also needs organic revenue growth of 8% or better in 2027, with Pratt's margin holding above 10%. The bear thesis is proven right if full-year free cash falls beneath $9 billion or engine shop expenses drag Pratt's margins back under 9%.

Our take: Accumulate. A 36% backlog surge and a debt rating upgrade rarely arrive alongside a 47% quarterly cash drop. That divergence provides a buying window, provided the fourth quarter closes the cash gap as scheduled.

3. Netflix (NFLX) — The Gym Membership

Netflix is the gym that spent years growing by signing up new members. Now it grows by lifting rates on the regulars who never cancel their autopay.

The company commands the world's leading paid streaming entertainment platform, serving roughly one billion viewers. Its engine runs on recurring consumer subscription dues, complemented by an entry-level tier that subsidizes cheaper monthly pricing by inserting programmatic video advertisements into popular series.

Operational execution remains clean. Third-quarter revenue advanced 11.8% to $12.87 billion, while operating margins expanded to 34.6% from 28.2% in the prior-year period. Diluted earnings per share surged 49% to $0.87, and free cash flow reached $2.88 billion. The company deployed $4.8 billion to retire 63.9 million shares—its largest buyback quarter ever—leaving $22.3 billion on its authorization board against gross corporate debt of $15.3 billion.

Management projected fourth-quarter revenue of $13.4 billion, pointing to full-year sales of roughly $51.1 billion with operating margins clearing 31.5%.

Despite the numbers, the stock sits roughly 46% below its record peak and about 23% below its 200-day average, lingering near 52-week lows. Growth has drifted down from mid-teens rates to about 12%. At the same time, advertising revenue targets for the year were cut from roughly $3 billion to $2.7 billion because some ad products rolled out slowly. That slip contributed to roughly 1,000 corporate layoffs in October. But the selloff compressed Netflix's multiple to roughly 16.5 times forward earnings, with a free cash flow yield near 4.5%. That is its lowest valuation in five years, against a stock that spent most of the past five years trading at about twice that multiple.

The short thesis is simple: customer viewing time is rolling over. Netflix executives have admitted that investors are scrutinizing viewing hours, and private analytics show total time spent on the app declining. When advertising revenue underperforms, the business must hike subscription prices to hit its numbers. Pushing price increases onto consumers who are already streaming fewer shows inevitably spikes churn. If subscriber counts drop, operating margins will crack under the pressure of non-negotiable studio budgets.

What we're watching: Fourth-quarter earnings on January 19, 2027. Keeping the recovery on track requires 2027 top-line guidance of at least 12%, an advertising revenue target above $3.5 billion, and operating margins projected at 31% or better. The bear case takes over if full-year revenue growth guides below 10% or if margins slip below 30% as production spending resumes.

Our take: Accumulate. This is a classic mispricing where sentiment has outrun reality. You are picking up an established global platform at its cheapest valuation in half a decade because an advertising side business missed its timeline. Meanwhile management is using record cash flow to take shares off the board.

4. Advanced Micro Devices (AMD) — The Second Source

For a decade AMD was the cheaper, almost-as-good option to Intel. Now it wants to be the cheaper, almost-as-good option to Nvidia, except the market prices "almost" as if it were identical.

AMD designs central processing units (EPYC) and graphics accelerators (Instinct) for corporate clouds, alongside PC microprocessors. It is now attempting to challenge Nvidia's complete computing stack with its Helios server architecture, packaging racks of processors, memory, and high-speed networking equipment directly into complete turnkey computing clusters.

The hardware momentum is undeniable. Third-quarter revenue jumped 45% to $13.4 billion, propelled by a 97% year-over-year surge in data-center revenue to $8.5 billion. Non-GAAP gross margin—the cash retained after paying third-party foundries for silicon wafers—held at 56%, lifting adjusted per-share earnings 74% to $2.09.

Data-center computing now drives 64% of company sales, easily offsetting an 8% slide in personal computing chips and a 40% collapse in video game console components. Management guided fourth-quarter sales to roughly $17 billion, while executives suggested on their conference call that 2027 revenue could jump more than 80% as data-center hardware takes over 80% of company shipments.

The shares jumped roughly 19% on the week, trading about 26% above their 200-day trendline with annual gains near 149%. Huang's departure from Nvidia directly fueled the run, as cloud operators searched for alternative silicon to hedge their procurement risks. That rotation leaves AMD trading near 30 times forward earnings, against roughly 18 times for Broadcom. It also carries an enterprise value near 78 times EBITDA and a price-to-free-cash-flow multiple above 100 times, a cash yield under 1%.

The risk sits inside the warehouse. Free cash flow shrank to $554 million during the quarter, down from $2.57 billion at the start of the year. In order to construct complete Helios server cabinets, AMD let inventories climb to $11.5 billion from $7.9 billion at the start of the year, while total debt climbed to $7.1 billion from $3.2 billion a year earlier.

AMD is spending real cash to manufacture enterprise hardware before receiving customer payments. If one hyperscaler delays installation of its planned server clusters, $11.5 billion of inventory faces write-downs. That would push the 2027 profit targets out while the stock still trades at twice the valuation of rival chipmakers.

What we're watching: Fourth-quarter earnings on February 3, 2027. Bullish investors need revenue of at least $17 billion, data-center sales topping $11 billion, gross margins holding at or above 55%, and formal 2027 guidance affirming the 80% trajectory mentioned on the call. The bear case is confirmed if Helios delivery dates slip, if full-year 2027 revenue growth guides under 60%, or if quarterly data-center revenues miss $10.5 billion.

Our take: Wait for a better entry. A near-doubling in data-center sales is exceptional industrial execution. But you are being asked to buy a cyclical semiconductor stock at peak multiples on the back of an informal comment made on an earnings call. We would look to build positions on a market retreat that pulls AMD's multiple closer to Broadcom's, or after Helios revenues convert directly into free cash rather than unsold warehouse stock.

The bottom line

When leadership changes rattle the tech sector and the Federal Reserve threatens to make capital expensive again, speculative financial projections stop working. Constellation gives you locked twenty-year energy agreements backed by the strongest corporate balance sheets in tech. RTX anchors you in commercial maintenance contracts alongside an expanding global defense backlog. Netflix offers a five-year valuation discount while its recurring subscriber base fuels historic share repurchases. AMD asks you to pay a record premium for warehouse hardware that has not yet converted to cash flow. We will track the incoming cash numbers against each commitment as the winter earnings cycle lands.

Analysis, not advice. We may own these names and we may sell them at any time.