CoreWeave Backlog Soars, Guidance Sinks Stock.

Mazen Al Sabbagh

— 7 min read
Share

TL;DR

  • CoreWeave reported Q1 revenue of $2.08 billion, exceeding expectations, but shares declined 10% post-market due to light Q2 guidance ($2.45-2.6B vs. $2.69B consensus) and a wider-than-expected loss of $1.40 per share. The company's revenue backlog reached $99.4 billion, while 2026 capex projections increased to $31-35 billion. (CNBC, Benzinga)
  • Cloudflare initiated a workforce reduction of 1,100 employees, representing 20% of its staff, to align with an "agentic AI-first" operating model, following a reported 600% surge in internal AI usage over three months. Despite a Q1 revenue beat, the stock experienced an 18% decline. (CNBC, Reuters)
  • During the November 2023 crisis, OpenAI's board actively considered a merger with Anthropic, with Dario Amodei as the proposed CEO, according to former board member Helen Toner's testimony in the ongoing trial. Toner characterized it as a viable option among difficult alternatives. (The Verge)
  • Nvidia committed up to $2.1 billion to IREN, a data center operator transitioning from bitcoin mining, to facilitate the deployment of up to 5 gigawatts of AI infrastructure, marking another strategic compute partnership in Nvidia's expanding portfolio. (Reuters, CNBC)
  • China's Moonshot AI, developer of the Kimi chatbot, completed a $2 billion funding round at a $20 billion valuation, led by Meituan's venture arm and China Mobile, bringing its total capital raised to $3.9 billion within six months amidst accelerated Chinese AI investment. (TechCrunch, Bloomberg)

Lead Story: CoreWeave Backlog Soars, Guidance Sinks Stock

CoreWeave's inaugural public quarterly earnings release on Thursday presented a bifurcated operational narrative. The company’s top-line performance exhibited robust growth, while forward guidance introduced significant investor concern.

First-quarter revenue reached $2.08 billion, a 52% year-over-year increase, comfortably surpassing the $1.97 billion Wall Street consensus. The period marked the company's "strongest bookings quarter in history," securing over $40 billion in new commitments and elevating its contracted revenue backlog to $99.4 billion. This backlog figure alone positions CoreWeave among the most substantially committed infrastructure providers globally. Concurrently, S&P upgraded the company's credit rating to positive. CEO Mike Intrator emphasized the company's strategic alignment, stating, "As the market moves from training to inference, that distinction matters more than ever. CoreWeave was built for exactly this."

Conversely, the bottom line deteriorated. CoreWeave reported a net loss of $740 million, translating to -$1.40 per share, a 54% wider loss than the -$0.92 analysts anticipated. The most significant market reaction stemmed from the Q2 revenue guidance, set at $2.45-2.6 billion, with the midpoint of $2.53 billion notably trailing the $2.69 billion LSEG consensus. While full-year 2026 revenue guidance remained at $12-13 billion, capital expenditures were revised upward to $31-35 billion from a previous floor of $30 billion. The company also increased its exit run-rate revenue target to $18-19 billion annualized.

Shares declined as much as 10% in after-hours trading. Following an 80% year-to-date rally, the stock had priced in sustained operational outperformance. The Q2 guidance discrepancy introduced questions regarding the velocity at which backlog converts into recognized revenue. Goldman Sachs concurrently disclosed an 8.1% passive stake via a Schedule 13G filing.

The broader implications for AI infrastructure are nuanced. CoreWeave's demand metrics are exceptional—nearly $100 billion in backlog and 3.5 gigawatts of contracted power. However, the delta between committed demand and near-term revenue recognition is wider than market expectations. Given Anthropic's $200 billion Google Cloud commitment and Nvidia's IREN deal, the fundamental appetite for AI compute remains undisputed. The critical question shifts to whether infrastructure providers can translate this demand into predictable, profitable revenue at the pace required by public market valuations. (CNBC, SiliconANGLE, TipRanks, Investing.com)

In Other News

Cloudflare Cuts 1,100 Jobs to "Architect for the Agentic AI Era." Cloudflare announced the elimination of over 1,100 positions, approximately 20% of its total workforce, in a restructuring explicitly attributed to an "AI-first" operational pivot. CEO Matthew Prince and co-founder Michelle Zatlyn's memo, "Building for the Agentic AI Era," cited a 600% surge in internal AI usage and thousands of daily AI agent sessions among employees within three months. The reductions span engineering, HR, finance, and marketing. Despite exceeding Q1 revenue estimates, Cloudflare's Q2 guidance lagged consensus, causing shares to fall 18%. This strategic framing underscores a significant shift: the layoffs are presented not as a response to underperformance, but as a proactive re-architecture driven by the perceived labor efficiency gains of agentic AI. (CNBC, Reuters, Business Insider, LA Times)

The Merger That Almost Was: OpenAI's Board Discussed Combining with Anthropic. The eighth day of Musk v. Altman yielded a critical revelation, recasting the November 2023 OpenAI crisis. Former board member Helen Toner testified that during the period dubbed "the Blip," the board engaged in serious discussions regarding a merger of OpenAI with Anthropic, considering Dario Amodei as the potential CEO of the combined entity. Toner affirmed this as "an option worth considering among our set of difficult options." This disclosure fundamentally alters the historical narrative, suggesting the board explored not merely a replacement CEO, but a structural consolidation with OpenAI's primary competitor, under leadership that had departed OpenAI over safety disagreements with Altman. Further testimony from former board member Tasha McCauley and former employee Rosie Campbell detailed concerns regarding Altman's leadership and a perceived shift from research and safety to commercialization. (The Verge, NBC Bay Area, India Today, Washington Post)

Nvidia Extends Its Infrastructure Empire with $2.1 Billion IREN Deal. Nvidia continued its strategy of leveraging equity investments to secure AI infrastructure capacity, announcing an agreement with IREN. The deal grants Nvidia a five-year right to acquire up to 30 million shares at $70 each, representing a potential $2.1 billion investment. In return, IREN commits to deploying up to 5 gigawatts of Nvidia-designed AI infrastructure. This partnership aligns with Nvidia's prior arrangements with CoreWeave and Nebius, establishing a network of infrastructure operators financially integrated with Nvidia's hardware roadmap. IREN, in its strategic pivot from bitcoin mining to AI data centers, also announced a $3.4 billion infrastructure commitment. Shares initially surged 27% post-market before stabilizing with approximately a 6% gain. Nvidia's operational model is clear: rather than directly constructing data centers, it cultivates a network of captive infrastructure partners through warrants and investment deals, guaranteeing the deployment of its chips at scale. (Reuters, CNBC, Nvidia Newsroom)

X / Social Pulse

CoreWeave's earnings dominated financial AI discourse on Thursday evening. Sentiment was dichotomous: infrastructure proponents emphasized the $99.4 billion backlog as proof of unyielding demand, while skeptics highlighted the widening losses and the $31-35 billion capex as a significant cash-flow management challenge demanding flawless execution. Several analysts underscored the Q2 guidance miss as the more salient indicator, noting that backlog represents commitment, whereas guidance reflects management's near-term revenue conversion expectations.

Cloudflare's layoffs generated a distinct segment of market anxiety. The term "agentic AI era" trended alongside CEO Prince's memo, fueling discussions less about Cloudflare's specific performance and more about the broader implications for white-collar employment. The reported 600% increase in Cloudflare's internal AI usage over three months is now a key data point for arguments concerning AI-driven labor displacement.

The revelation from Toner's testimony regarding a potential OpenAI-Anthropic merger elicited considerable astonishment from industry observers. The notion that two entities often positioned as ideological counterpoints in the AI safety debate were once days from combining under Amodei's leadership adds an unforeseen dimension to the Musk v. Altman trial, diverging from either legal narrative. Commentators speculated extensively on the counterfactual scenario: the profound reshaping of the AI industry had such a merger transpired.

One to Watch

The White House AI Policy Whiplash. The Trump administration's stance on AI regulation underwent rapid oscillation this week. On Wednesday, Kevin Hassett suggested the White House was exploring an executive order to review AI models "like an FDA drug." By Thursday evening, Politico reported a senior White House official had walked back this position, indicating a preference for "partnership" over "government regulation." This contradiction exposes internal policy friction following David Sacks's departure. The West Wing lacks a singular voice on AI policy, creating a vacuum filled by factions with divergent instincts. While the national security concerns—epitomized by the "FDA for AI" analogy—persist, the political imperative to avoid appearing anti-innovation pulls in an opposing direction. The eventual executive order, confirmed to be in development by multiple sources, will reveal whether the administration prioritizes pre-market approval or voluntary collaboration, thereby defining the regulatory landscape for frontier model releases through the 2026 midterms. (Politico, Bloomberg, Tom's Hardware)

Quick Hits

  • TSMC reported April 2026 revenue of NT$410.73 billion ($13.08 billion), a 17.5% year-over-year increase, contributing to a 29.9% cumulative revenue rise for 2026 through April, driven by sustained AI chip demand. (DigiTimes)
  • China's Moonshot AI secured $2 billion in funding at a $20 billion valuation, with Meituan's venture arm and China Mobile leading the round, bringing its total funding to $3.9 billion in six months amid the rising global prominence of its Kimi chatbot and K2.6 model. (TechCrunch, SCMP)
  • Lambda successfully obtained a $1 billion credit facility to scale its Nvidia-powered AI server fleet and expand its gigawatt-scale data centers, positioning itself for the "superintelligence race." (TechStartups)
  • RBC Capital Markets revised its S&P 500 year-end target to 7,900 from 7,750, citing "resilient earnings growth and continued strength in AI-linked sectors." (Reuters)
  • Connecticut enacted HB 5222, establishing a pilot program for independent verification of AI governance, marking the first state-level legislative framework for third-party AI auditing. (Transparency Coalition, PR Newswire)

The week concludes with a definitive portrayal of an industry operating at divergent speeds. Demand signals remain extraordinary—CoreWeave's $99.4 billion backlog, TSMC's 30% revenue growth, Nvidia's expanding infrastructure partnerships, and Moonshot's $20 billion valuation on early ARR. Yet, the operational reality presents increasing friction: CoreWeave's losses are expanding concurrently with its revenue, Cloudflare is shedding a fifth of its workforce to adapt, and the White House demonstrates an inability to maintain a coherent AI policy position for even 48 hours. The forthcoming week will bring the full implications of the April jobs report, anticipated testimony from Altman (likely around May 11), and continued analysis of the revelation that OpenAI and Anthropic were once a board vote away from corporate unification.

Sources

Lock in. M. mazen@thorterminal.com

Read more