Deirdre Bosa 3.5 185 ideas

Engineers are being incentivized ("tokenmaxxing") to maximize AI token usage, which may create artificial or performative demand signals. The market is already questioning hyperscaler demand claims as the AI trade stalls. If a material portion of reported AI demand is driven by these internal incentives rather than genuine utility, the foundational assumption behind massive capital expenditure (over $1 trillion) becomes shaky. The sector carries heightened risk of a negative reassessment as the quality of demand comes under scrutiny, making it an unattractive or "AVOID" area until visibility improves. Underlying enterprise adoption accelerates independently of the token incentives, validating the current spending plans and growth trajectory.

SpaceX's IPO is reportedly expected to allocate as much as 30% of shares to retail investors, far above the 5-10% norm, with Elon Musk aiming to build loyalty for control from the start, similar to Tesla. High retail allocation for loyalty rather than market clearing can lead to poor price discovery and performance, as evidenced by Saudi Aramco's IPO where retail got about a third of shares and the stock went sideways for years. AVOID because the IPO may be overpriced or prioritize Musk's control over investment fundamentals, increasing risk for public market investors. If SpaceX delivers strong operational performance or retail demand is based on genuine growth prospects, the avoidance thesis could break.

Apple is blocking AI coding apps (e.g., Replit) from updating in the App Store, citing safety concerns, despite similar allowances for other apps like Facebook and Claude. This inconsistent policy risks alienating software builders who may bypass the App Store for web development, avoiding Apple's review and 30% cut, while Apple is already late to AI adoption. Apple could become irrelevant to the next generation of AI builders, harming its ecosystem growth and long-term revenue potential in a critical technological shift. Apple might resolve internal confusion, adjust policies to embrace vibe coding, or leverage the trend for monetization, mitigating the strategic risk.

We just got our CNBC Mac Mini. We're still really thinking about how we set it up. We're not going to let it touch anything in the organization... I think Nvidia at GTC is even going to have its own way of setting up your own open claw. Enterprises are terrified of data leaks from cloud-based AI. The solution is local, self-hosted, air-gapped AI models. To run these highly capable models locally and securely, enterprises will need to purchase massive amounts of on-premise Nvidia hardware rather than just renting cloud compute. LONG. The shift toward secure, self-hosted enterprise AI creates a massive new hardware upgrade cycle for Nvidia outside of traditional hyperscalers. Apple or AMD capture the local/edge compute market, or enterprises decide cloud security is sufficient, reducing the need for on-premise hardware.

The information reported a potential tie up between private equity and anthropic, in this case, the Blackstone with Anthropic... private equity is very different than like each individual enterprise deciding whether or not to use AI. I mean it can just be mandated from the top down. If Blackstone secures a strategic partnership with the leading enterprise AI model, it can force rapid AI adoption across its massive portfolio of non-tech companies (retail, hospitals, logistics). This top-down mandate will drastically cut operational costs and boost margins across Blackstone's holdings, increasing the firm's overall returns. LONG. Blackstone gains a unique structural advantage in modernizing legacy businesses through forced AI integration. The rumored tie-up falls through, or portfolio companies face severe implementation hurdles and security risks when deploying AI at scale.

Anthropic is in talks with Blackstone and other firms to form an AI joint venture embedding Cloud across their portfolio companies. Blackstone can use its board control to force portfolio companies to adopt AI, drastically reducing their operational costs. This margin expansion directly boosts Blackstone's fund IRR, making their funds more attractive to LPs, driving higher performance fees, and increasing the firm's overall valuation. LONG BX as a primary beneficiary of AI-driven cost optimization in the private markets. Implementation costs of AI may offset initial savings, or AI models may not be reliable enough to fully replace specialized human workflows.

Say a Blackstone company uses Cloud to replace a bunch of point solutions... Nobody cancels Salesforce because they saw a cool demo, but PE firms, they have board control, IRR targets and a ticking clock. Traditional enterprise SaaS relies on high switching costs and sticky recurring revenue. However, PE-owned companies are highly incentivized to cut costs and have the top-down authority to rip out expensive software seats (like Smartsheet or Salesforce) in favor of cheaper, bundled AI agents. This will lead to unexpected churn and revenue destruction for legacy SaaS providers. AVOID legacy enterprise SaaS companies, as they face unprecedented churn risks from cost-conscious, PE-backed clients. AI agents may fail to replicate the complex, compliant workflows of established SaaS platforms, forcing companies to maintain their traditional software subscriptions.

ServiceNow, Datadog, and Shopify have all bounced back from the severe software sell-off. The market is picking winners in that sell-off and CRM just isn't one of them yet. Investors are actively rotating capital out of legacy software companies relying on financial engineering and into high-quality names that are successfully maintaining organic growth and navigating the AI transition. These relative strength leaders will continue to capture premium multiples as capital concentrates in proven winners. LONG. These companies have proven their resilience and are being rewarded by the market as structural leaders in the enterprise software space. A broader macroeconomic slowdown or a sector-wide multiple compression could disproportionately impact these higher-valuation growth names regardless of their individual execution.

Salesforce is levering up and taking on debt to buy back stock, which is the playbook of IBM and old Oracle, while paying about $3.5 billion a year in stock comp. Borrowing heavily to fund buybacks rather than investing in AI infrastructure or M&A signals that management may lack high-return internal growth opportunities. Furthermore, a significant portion of the buyback simply absorbs stock-based compensation dilution rather than creating true shareholder value, leaving the company vulnerable if its core growth slows. AVOID. Tripling the company's debt to engineer earnings per share, while organic AI revenue remains under 2 percent, presents an unfavorable risk profile compared to faster-growing software peers. If AI disruption fears are overblown and Agentforce scales rapidly, the leveraged buyback will result in massive EPS accretion and drive the stock significantly higher.

"Others like Amazon and Oracle, even Alphabet, they are borrowing but they're not borrowing for buybacks. They're borrowing to fund growth. Many of them are spending on infrastructure which will bolster the AI bet." Companies that leverage debt to build out capital-intensive AI infrastructure are creating deep competitive moats and securing future revenue streams. This productive use of debt positions them to dominate the next technological era, unlike companies borrowing purely for financial engineering. LONG AMZN, ORCL, and GOOGL as their debt-funded infrastructure investments will drive long-term AI leadership. The return on investment for AI infrastructure takes longer than expected to materialize, leading to bloated balance sheets and investor impatience.

"Salesforce, the poster child of the software sell off, tripling its debt to prop up the stock price is a gamble that AI disruption fears are overblown... Agent Force is actually still less than 2% of total revenue." Borrowing heavily to fund stock buybacks instead of investing in AI infrastructure or M&A signals that management lacks organic growth ideas. Furthermore, these buybacks largely serve to mask the dilution from $3.5 billion in annual stock-based compensation, making the company look more like a legacy tech stock (e.g., old IBM) than an AI pioneer. AVOID CRM as its reliance on financial engineering fails to address underlying concerns about AI disruption and slowing core growth. Management's conviction is correct and the stock is truly undervalued, or Agentforce adoption accelerates enough to meaningfully impact the top line.

"While other software names like ServiceNow, Datadog, Shopify, well, they have bounced this month, Salesforce is flat, so the market is starting to pick favorites in the SaaS selloff." Investors are actively rotating capital within the software sector, moving away from legacy names struggling with growth and into resilient SaaS companies that are successfully navigating the AI transition. LONG NOW, DDOG, and SHOP as they are emerging as the market's preferred winners in the current software environment. A broader macroeconomic downturn or a sector-wide SaaS multiple compression could drag down these names regardless of their individual outperformance.