☀️ Morning Briefing
Highlighted Companies
KEY COMPANIESDespite Agentforce’s $800M ARR milestone, Benioff is publicly countering the AI job-destruction narrative by committing to entry-level hiring — a deliberate positioning move as CRM defends its talent and customer perception against enterprise AI competitors. The signal matters: Salesforce is framing Agentforce as augmentation, not replacement, which directly shapes how enterprise procurement committees evaluate the platform’s political risk.
Read More →NOW beat on topline ($3.77B, +22% YoY) and raised full-year guidance to $15.755B — yet shares collapsed 14% after management disclosed on-premise deal slippage tied to the Iran conflict, with several large MENA contracts pushed out of Q1. The market is discounting the geopolitical exposure heavily, even as AI platform growth (16 deals over $5M in net new ACV, +80% YoY) remains structurally strong.
Read More →Google’s Agent2Agent (A2A) protocol — enabling cross-vendor agent handoffs between Agentforce, Copilot Studio, and Gemini Enterprise — now has 150 organizations live in production. This is the emerging plumbing layer of the enterprise AI stack, and Google’s push to become the neutral orchestration layer directly challenges both Salesforce’s and ServiceNow’s bids to own the agentic OS.
Read More →The disconnect between Agentforce’s operational momentum (29K deals, 169% YoY ARR growth) and CRM’s equity price (-31% YTD) reflects a genuine market debate: does multi-vendor agentic interoperability erode Salesforce’s CRM moat, or does $800M ARR prove the opposite? At current prices, the stock trades at a meaningful discount to ServiceNow on forward multiples, creating a potential entry point for long-horizon investors.
Read More →As enterprises accumulate agents from Copilot, Gemini, and Agentforce simultaneously, both Salesforce and ServiceNow are positioning as the centralized orchestration and governance layer — the OS of the multi-agent enterprise. Whoever wins this architectural role captures a durable revenue layer independent of which underlying model wins. The battle is as much about workflow data ownership as it is about AI capability.
Read More →AI & Semiconductors
AI & CHIPSIntel reported Q1 EPS of $0.29 adjusted vs. $0.01 expected on $13.58B revenue ($12.42B est.) — the strongest earnings beat of the semiconductor cycle so far. Data Center & AI grew 22% to $5.1B as Google committed to multiple future Intel CPU generations for AI workloads. With Q2 guidance of $13.8–$14.8B well above the $13.07B consensus, Intel’s turnaround narrative just gained significant credibility.
Read More →Broadcom signed expanded agreements to supply AI chips to both Google (future TPU generations) and Anthropic (~3.5 gigawatts of compute capacity). Mizuho estimates Broadcom picks up $21B in AI revenue in 2026 and $42B in 2027 from these deals alone. This cements Broadcom as the dominant custom ASIC supplier — and the clearest structural beneficiary of hyperscalers moving away from general-purpose GPU procurement.
Read More →Meta’s expanded Broadcom commitment (1GW of custom silicon) comes alongside Hock Tan agreeing to leave Broadcom’s board — a governance clean-up that likely preceded the deal’s announcement. Meta is simultaneously deploying AMD GPUs (up to 6GW), Nvidia chips, and Arm-designed custom silicon, executing a deliberate multi-vendor diversification strategy to reduce single-supplier AI infrastructure risk.
Read More →Anthropic is exploring custom silicon to manage compute costs for Claude inference — mirroring moves by Amazon (Trainium/Inferentia), Google (TPU), and Microsoft (Maia). The Motley Fool analysis argues the real Nvidia threat is not AMD or Broadcom but the hyperscalers’ and frontier AI labs’ own silicon programs collectively displacing GPU purchases at the margin. Nvidia’s $1T Blackwell/Vera Rubin projection still holds, but custom ASIC share is structurally growing.
Read More →OpenAI’s co-designed chip with Broadcom has slipped from its original Q2 target to Q3 2026, keeping the company firmly in Nvidia’s Blackwell ecosystem through at least year-end. The delay is notable given CFO Sarah Friar’s concerns about compute contract affordability — if custom silicon can’t come online faster, OpenAI’s unit economics on inference remain dependent on Nvidia’s pricing power.
Read More →Markets & Tech Stocks
S&P 500 · NASDAQ · MARKETSReports surfaced that Friar has warned company leadership that revenue growth may not support OpenAI’s existing compute contracts — a direct challenge to Altman’s capex-first strategy. Friar separately stated OpenAI won’t be IPO-ready by end of 2026. Despite a joint statement calling the reports “ridiculous,” the market read was clear: OpenAI revenue miss triggered a sell-off in AI infrastructure stocks and raised broader questions about demand absorption across the AI capex build-out.
Read More →The S&P 500 rose to 7,165 and the Nasdaq to 24,837 on April 24, with the rally led by semiconductor and mega-cap AI names. Nvidia retaking the $5T market cap mark and Intel’s 94% YTD run signal that the market is pricing an AI infrastructure supercycle as durable, not cyclical — even as OpenAI’s demand signal introduces the first meaningful demand-side caution of the year.
Read More →FactSet’s April 24 update shows 84% of S&P 500 reporters beating EPS estimates with earnings 12.3% above forecasts — a beat rate and magnitude both above 5-year averages. At 16.1% aggregate YoY growth, the earnings picture is materially better than the 11-12% analysts expected entering Q1. The macro tailwind is real, but the market’s attention is now shifting to forward guidance quality as AI capex commitments weigh on FCF projections.
Read More →The divergence between NOW’s operational raise (full-year sub-rev guidance up to $15.755B midpoint, ~21% CC growth) and its -14% single-day move is a textbook case of market over-discounting a one-quarter geographic disruption. RPO of $27.7B (+25% YoY) and cRPO of $12.64B (+22.5% YoY) both suggest the demand pipeline is healthy outside the MENA disruption. Investors willing to look past Q1 noise are getting a discount on a compounding AI platform.
Read More →With Apple, Microsoft, and Meta all reporting this week, markets entered Monday on rangebound caution despite the prior week’s record closes. The focus will be on AI monetization specificity: investors want to see whether Copilot and Azure AI are converting to measurable ARR, and whether Meta’s AI capex is producing measurable user engagement returns — or whether the OpenAI demand-miss is a canary for broader AI ROI skepticism.
Read More →Supply Chain & Commodities
CHIPS · MATERIALS · FREIGHTHBM’s 3:1 wafer displacement ratio against DDR5 is now structurally compressing general-purpose memory supply. With SK Hynix and Micron advanced packaging lines at capacity through 2026 and Micron’s new Idaho fab not online until 2027, the 80–90% QoQ price surge in server DRAM is not transitory — it’s the consequence of a deliberate allocation framework that prioritizes hyperscalers over OEMs. Enterprise server refresh cycles will feel this through at least mid-2027.
Read More →The top eight cloud providers are on track to deploy over $600B in capex in 2026 — a 40% YoY acceleration — with Project Stargate alone committing Samsung and SK Hynix to up to 900,000 DRAM wafer starts per month. At this scale, the bottleneck is no longer just chips: advanced packaging capacity, substrate materials, power infrastructure, and skilled fab labor are all constraint surfaces. The supply chain is running hot with no near-term relief valve.
Read More →Despite record AI demand, chipmakers are deliberately holding back capacity expansion — a lesson learned from the 2022-23 oversupply cycle. The result is a tighter, more disciplined supply chain that benefits incumbents (TSMC, SK Hynix, Samsung) at the expense of new entrants and legacy OEMs. Micron’s Hiroshima HBM facility won’t begin construction until May 2026, with first output in 2028 — confirming that the current supply squeeze has years to run.
Read More →The OpenAI revenue-miss story hit AI infrastructure names hardest — CoreWeave, Oracle (cloud), and data center REITs all pulled back on concerns that AI workload demand growth may be softening relative to supply build-out commitments. The risk is a demand-supply mismatch: hyperscalers have locked in $600B+ in capex based on AI adoption curves that may be running 6–12 months behind internal targets. Friar’s concerns about compute contract affordability are the clearest public signal yet of this tension.
Read More →Chips from AMD, Broadcom, and Intel have all been selected for deployments by OpenAI, Anthropic, Meta, Microsoft, and Amazon — signaling that the AI infrastructure buildout is large enough to sustain multiple winning vendors simultaneously. Nvidia’s $1T Blackwell/Vera Rubin projection remains intact, but the growing custom ASIC share means supply chain diversification is structurally increasing, reducing single-vendor concentration risk for hyperscalers over a 3–5 year horizon.
Read More →