• Quanta Services (PWR) stands out as the clearest actionable infrastructure opportunity: EPS rose 82%, organic revenue grew 37%, backlog hit a record $64 billion, and the company raised guidance.
• GE Vernova (GEV) offers exposure to rising power demand, with orders up 96% to $29 billion and EPS up 193%; be mindful that the stock fell despite strong results.
• Treat CoreWeave (CRWV), Nebius (NBIS), and IREN cautiously: high borrowing costs and power-access constraints could derail data-center growth.
• Approach the potential Anthropic and OpenAI IPOs cautiously; scrutinize growth, pricing, compute costs, and profitability rather than relying on AI adoption alone.

Anthropic (Private)

• Anthropic’s S-1 showed $6.2 billion in debt, a 19% adjusted operating margin in Q2 2026, and revenue concentrated to a limited degree: its five largest customers accounted for 25% of first-half revenue.
• The speaker highlighted approximately $800 billion in future compute and equipment commitments listed in the filing, including commitments involving Microsoft, Google, Amazon, SpaceX, AMD, Broadcom, NScale, and Lambda. He warned that these commitments leave little room for error and could drive compute costs sharply higher from 2027.
• Revenue growth slowed after the end of “token maxing,” when companies had encouraged broad AI use with less concern about cost. Sequential revenue growth fell to 45% in Q3 2026, from 147% in Q2. The revenue-to-compute-expense ratio also declined, from 2.08x to 1.70x.
• Anthropic cut token prices by as much as 50% in October to stimulate demand. The speaker questioned whether increased usage would offset the lower prices and said the company’s path to profitability remains uncertain.
• The speaker expects Anthropic’s IPO to proceed and said OpenAI may follow, but did not provide a valuation or investment recommendation.

Takeaways

• Treat the potential IPO as a high-uncertainty opportunity: the discussion raised concerns about slowing growth, pricing pressure, substantial infrastructure commitments, and profitability.
• Watch revenue growth and compute costs together. Continued adoption alone may not translate into attractive margins if prices fall or infrastructure spending rises.

OpenAI (Private)

• OpenAI was described as likely to follow Anthropic with an IPO, but the transcript offered no financial figures or specific view on its valuation.
• The speaker grouped OpenAI with Anthropic as a company that may have benefited from “token maxing,” which he believes ended as enterprise customers became more sensitive to AI costs.

Takeaways

• The transcript provides no basis for estimating OpenAI’s value or prospects. If it comes to market, the speaker’s comments suggest scrutinizing customer demand, pricing, and compute costs.

AI Infrastructure and Data-Center CapEx

• The speaker questioned whether the AI capital-spending cycle is slowing after weaker-than-expected signals from Microsoft and Amazon, alongside negative market reactions across AI-related stocks.
• An electrical contractor quoted in the episode said power availability, not GPUs, memory, or financing, could be the main constraint on data-center construction. Transmission lines may take 24 months or more, and substations may take one to two years.
• The contractor warned that announced investment can become “stranded CapEx” if data centers cannot get connected to power. He expected this constraint could slow construction in 2027 or later.
• The speaker said Microsoft and Google appeared better positioned on power planning, while Amazon and Meta could feel more pressure. He characterized CoreWeave and Oracle as further back in the queue for power.

Takeaways

• AI infrastructure exposure is not just a bet on demand for computing. The discussion highlights power connections, construction timing, and the ability to put announced capacity into service as important factors.
• Distinguish companies with operating infrastructure and access to power from those whose growth depends on projects that may be delayed.

Microsoft (MSFT)

• Microsoft reported 19% revenue growth, 16% earnings-per-share growth, and 48% Azure revenue growth. Its results were described as strong.
• The concern was capital spending: Microsoft reported $45 billion in CapEx and finance leases, below the $51 billion consensus. The news contributed to declines in the “neo-cloud” group, including CoreWeave, Nebius, and IREN.
• The speaker’s contractor source considered Microsoft relatively well positioned on data-center power planning.

Takeaways

• The discussion presents Microsoft as having strong operating momentum, especially in Azure, but also makes its spending plans and data-center capacity important to monitor.
• Strong cloud growth does not eliminate execution risk if infrastructure spending or power availability fails to keep pace with demand.

Alphabet / Google (GOOGL, GOOG)

• Revenue grew 25%, but free cash flow was negative $9.1 billion, worse than expected. Search revenue growth slowed to 12%, from 19% in Q1 2026.
• The speaker viewed the slower search growth as a sign that AI search may be gradually displacing Google’s traditional search business. He argued that conventional search is a high-margin, capital-light business, while AI search requires much heavier investment and may have lower barriers to customer switching.
• Google was also among the companies the contractor source viewed as better positioned to manage data-center power planning.

Takeaways

• The central issue raised was not simply Google’s AI spending, but whether AI could weaken its highly profitable search business while requiring more capital-intensive infrastructure.
• Track search growth, free cash flow, and the returns on AI investment alongside Google’s cloud and AI progress.

Amazon (AMZN)

• Amazon reported $202 billion in revenue, up 12% but below expectations. Earnings per share of $1.89 also missed estimates and were down from $1.95 a year earlier.
• Amazon kept its 2026 CapEx budget unchanged. The speaker said this was notable because hyperscalers had previously been raising their budgets quarter after quarter, and it could prompt investors to question whether AI CapEx growth is ending.
• The contractor source expected Amazon to feel power-availability constraints more than Microsoft and Google.

Takeaways

• Monitor whether Amazon’s flat CapEx budget reflects a temporary pause, spending discipline, or a broader slowdown in AI infrastructure investment.
• The discussion also flags power access as a possible constraint on Amazon’s ability to turn planned spending into operational data-center capacity.

Meta Platforms (META)

• Meta had more than $800 billion in AI-related commitments, according to the speaker. To fund its commitments, it announced a $95 billion equity raise, with $35 billion to be raised immediately and the remainder later.
• NVIDIA announced it would buy $3 billion of the offering as a show of support.
• The contractor source suggested Meta could feel power constraints more than Microsoft and Google.

Takeaways

• Meta’s AI plans are large, but the transcript raises questions about how much capital they require and whether the company can secure the power needed to support the buildout.
• Assess the spending alongside its funding needs and the pace at which new data-center capacity becomes usable.

NVIDIA (NVDA)

• NVIDIA said it would buy $3 billion of Meta’s equity offering, which the speaker characterized as a sign of support.
• NVIDIA was also mentioned as part of the broader AI hardware and data-center complex potentially affected if AI CapEx slows.

Takeaways

• The episode did not provide a specific valuation view on NVIDIA. Its actionable theme was to watch whether hyperscaler spending and data-center construction continue to support demand for AI hardware.

CoreWeave (CRWV)

• CoreWeave’s debt costs were a major concern. The speaker said its bonds’ effective yield had risen from 9% to 15%, reflecting lower bond prices and perceived default risk; the company still pays the stated 9% coupon.
• CoreWeave had $36 billion in total debt and slightly more than $50 billion of debt, lease liabilities, and other financial obligations combined.
• The speaker compared CoreWeave’s model to specialty finance: it borrows in fixed-income markets, builds data centers, and rents compute capacity. He warned that it must continue to raise debt to expand, potentially at much higher costs.
• He also highlighted CoreWeave’s competition with hyperscalers that have much larger balance sheets and other businesses. The contractor source said CoreWeave was far back in the queue for power connections.

Takeaways

• The transcript presents CoreWeave as a particularly risky way to invest in AI infrastructure because its growth depends on both debt-market access and successful data-center construction.
• Monitor borrowing costs, refinancing needs, lease obligations, and power availability—not just demand for rented compute.

Nebius (NBIS) and IREN (IREN)

• Nebius and IREN were identified as part of the “neo-cloud” group that fell 5% to 9% after Microsoft reported lower-than-expected CapEx.
• The speaker noted that the group was already under pressure from higher debt costs.

Takeaways

• These companies were presented as exposed to changes in hyperscaler spending and financing conditions. The transcript offers no company-specific financial analysis, so it supports caution rather than a firm view on either stock.

Oracle (ORCL)

• Oracle was named as a company that, according to the contractor source, was far back in the queue for power connections to data centers.
• The discussion did not provide Oracle earnings or a specific stock assessment.

Takeaways

• For Oracle’s AI infrastructure plans, the transcript highlights power access and construction timing as risks to monitor. It does not provide enough company-specific information to assess the stock more broadly.

AMD (AMD), Broadcom (AVGO), and AI Compute Suppliers

• AMD and Broadcom appeared among the companies named in Anthropic’s future compute and equipment commitments. The transcript listed $20 billion for AMD and $161 billion for Broadcom.
• These suppliers were also part of the broader hardware complex that could be affected if AI CapEx growth slows.
• The speaker cautioned that data-center power—not necessarily chip availability—could be the binding constraint on some projects.

Takeaways

• The discussion identifies these companies as potential beneficiaries of AI infrastructure spending but also highlights their exposure to a slowdown in spending or delays in data-center deployment.
• The transcript gives no stock-specific valuation or recommendation.

GE Vernova (GEV)

• GE Vernova makes gas turbines used to generate electricity and was described as a beneficiary of AI data-center power demand.
• Earnings per share rose 193% year over year to $4.81, and orders rose 96% to $29 billion.
• The stock fell on the earnings day despite strong results; the speaker suggested this could reflect a selloff after strong prior performance.

Takeaways

• The episode presents GE Vernova as an indirect AI-infrastructure opportunity through electricity generation, with strong reported operating results.
• The speaker’s caveat is that even strong results may not prevent a stock decline after a substantial run-up.

Quanta Services (PWR)

• Quanta builds power plants and other infrastructure for utilities and was described as a beneficiary of AI-related construction.
• Earnings per share rose 82% to $6.06, revenue increased 37% organically to $12 billion, and backlog reached a record $64 billion. The company also raised guidance, and its stock rose 5%.

Takeaways

• Quanta was one of the episode’s clearest positive examples of a company benefiting from infrastructure demand, including the need to build and connect power capacity.
• Its backlog and guidance were strong in the period discussed, though the transcript did not give a valuation or price target.

Caterpillar (CAT)

• Caterpillar was described as partly exposed to AI infrastructure through generators, turbines, and construction machinery.
• Revenue rose 17% to $20.68 billion and earnings per share rose 45% to $7.20, but both only modestly exceeded expectations. The stock fell 9% amid investor concern about AI infrastructure construction.

Takeaways

• The discussion shows that strong year-over-year growth may not be enough to satisfy investors if results only slightly beat expectations.
• The stock’s sensitivity to expectations about AI construction was a key risk raised in the episode.

Intel (INTC)

• Intel’s earnings per share rose 126% to $0.52, and revenue slightly beat expectations, but the stock fell 18% after hours.
• The company forecast that the PC slump would persist through at least the first half of 2027, with PC declines of 12% in 2026 and 10% in 2027. It also cited supply-chain problems in its data-center component business.
• Intel forecast fourth-quarter revenue of $15 billion to $16 billion, below the $17 billion consensus, and earnings per share of $0.25, versus a $0.43 consensus.

Takeaways

• The transcript emphasizes forward guidance over backward-looking earnings: Intel beat the reported quarter but gave a weaker outlook.
• Watch PC demand, supply-chain performance, and the gap between Intel’s data-center expectations and the broader AI investment narrative.

ServiceNow (NOW)

• ServiceNow’s third-quarter revenue rose 21% to $4.1 billion, while earnings per share increased 4% to $1.00, a slight miss.
• The main concern was guidance for current remaining performance obligations, or backlog: expected growth of 18.5%, below the 19.7% consensus and 21% in Q3. The speaker said this suggested a sharp deceleration in future revenue.
• The stock fell 27%, reviving concerns about the “SaaSpocalypse”—the risk that AI could undermine software companies.

Takeaways

• The episode highlights backlog growth and forward guidance as important indicators for software businesses, not just current revenue growth.
• The broader risk raised is that investors may reassess software companies if AI changes demand or slows future growth, even if the companies are deeply embedded in enterprises.

S&P Global (SPGI) and Moody’s (MCO)

• S&P Global reported 7% revenue growth, while its ratings business, referred to in the transcript as “Ratings,” grew 9%. The speaker said S&P and Moody’s form a duopoly but are not immune to interest rates.
• The stocks fell as concerns about AI’s impact on software companies resurfaced; both companies have software businesses.

Takeaways

• The discussion recognizes the strength of the ratings duopoly while flagging interest rates and AI-related software concerns as factors to monitor.
• No specific stock recommendation or valuation assessment was given.

Boeing (BA)

• Boeing reported a loss of $0.34 per share, an improvement from a loss of $7.47, but worse than the expected loss of $0.21. Revenue increased only 4%.
• Defense revenue rose 15% to $8 billion, while commercial aircraft revenue fell 3% to $10.7 billion.
• The stock fell 10% after the company missed expectations and provided disappointing cash-flow guidance. The speaker said Boeing’s turnaround remained incomplete.

Takeaways

• The episode portrays Boeing as a continuing turnaround story with mixed results and weak cash-flow guidance.
• Monitor execution in commercial aircraft and cash generation; the transcript identifies both as areas of concern.

Visa (V) and Mastercard (MA)

• Visa reported earnings per share of $3.48, up 17% and slightly above expectations. Payment volume rose 9%.
• Mastercard also reported good results, though not as strong as Visa; its dollar payment volume rose 7%.
• The speaker used payment volume as an indicator of consumer health, saying the figures showed that higher-income consumers were still supporting overall spending.

Takeaways

• Payment volume growth was the key positive signal for both companies in the discussion, with Visa’s 9% growth stronger than Mastercard’s 7%.
• The episode’s consumer takeaway is that aggregate spending remains supported by the top end of the market, even as lower-income consumers struggle.

Eli Lilly (LLY)

• Eli Lilly was described as having won the weight-loss drug competition.
• Third-quarter earnings per share were $7.48, up 7% and above expectations. Revenue rose 39% to $24 billion, and the stock gained 7%.

Takeaways

• The transcript highlights strong revenue growth and a leading position in weight-loss drugs.
• It provides no discussion of valuation, competitive risks, or a price target.

U.S. Telecom Companies: T-Mobile (TMUS), AT&T (T), and Verizon (VZ)

• T-Mobile reported earnings per share of $3.22, up 34%, and added 274,000 accounts, above expectations but below the 396,000 added a year earlier.
• The speaker said telecom stocks were under pressure because Elon Musk had stated that SpaceX’s Starlink could seek to become a fourth U.S. mobile provider.
• A subscriber argued that a satellite network capable of direct-to-phone service would be expensive, would require many low-orbit satellites that do not last long, and would mainly benefit a limited number of customers in remote areas. The speaker said terrestrial towers may be cheaper, but added that renting tower space could also be prohibitively expensive.
• The speaker expected the “telepocalypse” narrative to gain momentum as Musk announced more plans, while acknowledging that the satellite threat could be costly to execute.

Takeaways

• The transcript identifies a competitive threat to T-Mobile, AT&T, and Verizon, but also highlights substantial cost and coverage challenges for a satellite-based rival.
• Separate the impact of headlines from a functioning competing network: the discussion presents a credible potential risk, not a completed change to the market.

SpaceX / Starlink (Private)

• Elon Musk’s stated goal of turning Starlink from a satellite phone provider into a fourth U.S. mobile provider was described as a risk to the established telecom companies.
• The episode also noted that the network could require substantial spending on satellites and possibly tower access. The speaker said high costs had not historically stopped Musk from pursuing ambitious plans.

Takeaways

• The opportunity is speculative in the transcript: Starlink’s potential expansion could disrupt telecom, but its cost, network requirements, and ability to compete with terrestrial providers remain unresolved.
• No public ticker, investment terms, or timeline for the proposed mobile service was provided.

France Sovereign Debt

• The speaker argued that French sovereign debt is too small to pose a major threat to global fixed-income markets, comparing $4 trillion of French debt with $40 trillion of U.S. sovereign debt and BlackRock’s $15 trillion in assets under management.
• He said France needs the liquidity of markets such as BlackRock’s more than those markets need to own French debt, and concluded that investors do not need to own French sovereign debt.
• He also called political suggestions that France need not repay its debt “ridiculous.”

Takeaways

• The speaker’s view was bearish on French sovereign debt and that its global-market significance is limited by its size.
• This was a direct negative view, not a discussion of a specific bond, yield, or investment timeline.

BlackRock (BLK)

• BlackRock was mentioned as a scale comparison: it manages approximately $15 trillion, more than the amount of French sovereign debt cited in the episode.
• The speaker used this comparison to argue that France is not large enough to threaten the global fixed-income market.

Takeaways

• BlackRock was not discussed as a stock investment opportunity; the reference served as context for the speaker’s view on French debt.