CoreWeave's Squeeze Isn't Over Yet
1.16K Followers
Summary
- CoreWeave has surged 340% since its March 27 IPO, with short interest still high at ~32% and borrow rates peaking at 343%.
- Roughly 83.7% of shares are under a 180-day lock-up expiring in late September, posing a potential insider supply overhang.
- Q2 earnings in early August could trigger early lock-up sales if share conditions are met, adding near-term volatility risk.
- CoreWeave reported $982M in Q1 2025 revenue (+420% YoY) and guided to ~$5B for FY25, backed by a $25.9B backlog.
- Key growth drivers include $16B+ in OpenAI contracts, a new deal with Google Cloud, and Blackwell GPU deployment leadership.
CoreWeave, Inc.’s (NASDAQ:CRWV) 340% surge since its March 27 IPO, fueled by AI enthusiasm and a violent short squeeze, has turned heads across Wall Street. Our bullish call delivered a 289% return since my first bullish call and 58% since the second.
But as borrowing cost by short sellers decreases from its 343% peak and short interest stays high around 32%, the narrative is shifting from temporary technical catalysts to long-term fundamentals.
Investors should closely monitor two key dates: the early August release of the Q2 earnings, which could trigger the early sales of lock-up shares, and the end of the full 180-day lock-up in late September.
Lock-Up Expiration and Insider Supply Overhang
A major consideration for CoreWeave is the impending expiration of IPO lock-up agreements. Approximately 83.7% of outstanding shares (held by directors, executives, pre-IPO investors, etc.) are subject to a 180-day lock-up from March 27, 2025, prospectus date, with an early release trigger shortly after the company’s Q2 2025 earnings announcement.
In practical terms, the standard lock-up is set to lapse around late September 2025 (or two trading days after Q2 earnings, if sooner). Notably, CoreWeave’s lock-up terms include a provision that if the stock trades 25% above the $40 IPO price (>$50) for at least 5 out of 10 days after the first post-IPO earnings release (Q1 2025), roughly 1.0 million employee-held shares can be sold immediately. Given the stock indeed surpassed $50 in April, a small portion of insider shares likely became eligible for sale early.
The full lock-up expiration in Q3 poses a significant supply risk. Early investors and insiders who have seen multifold gains may look to realize profits, potentially flooding the market with shares. CoreWeave’s S-1 explicitly warns that when lock-up restrictions end, a wave of insider selling (or even the perception of it) could pressure the stock price.
The IPO included only ~37.5 million new Class A shares (plus ~0.9 million from a few selling stockholders, meaning the free float has so far been relatively limited. Post-September, hundreds of millions of currently-restricted shares (including those held by co-founders and venture backers) will become eligible to trade.
This time frame is key for investors: all CoreWeave holds must factor in volatility risk at the lock-up expiry date since insider selling may temporarily depress the stock price and drive trading volume sharply higher. Conversely, the lock-up expiry should enhance liquidity and new institutional buying but probably at the expense of weak short-term price action.
Short Interest and Technical Sentiment
CoreWeave’s meteoric rise has been exacerbated by extraordinarily high short interest and the potential for a short squeeze. In the weeks after the IPO, short sellers piled into the stock, perhaps questioning its rich valuation.
By late April, an estimated 45% of the public float was sold short (up from ~18% shortly after listing). This short interest continued to climb into late Q2. As of mid-June, reported short interest ranged from ~32% up to 45% of free float, with some speculative reports even claiming >60% (though such figures may double-count intraday churn). Even as the stock doubled, short selling continued. As of mid-June, around 32% of available shares were still on loan to short sellers, an extreme figure compared to names like Tesla (3%).
But the real pain came from the borrowing cost, the price short sellers pay to hold their positions. While borrowing costs for mega-cap stocks like Apple or Tesla are usually under 1%, CoreWeave’s borrow fees exploded past 150% annualized in June, peaking near 343% at one point. These fees were driven by sheer scarcity: at one point, just 25,000 shares were left to borrow market-wide. This made short positions not only risky but incredibly expensive to maintain.
The result? A classic short squeeze. As the stock rose and losses mounted, more than $1.6 billion in paper losses for short sellers by mid-June, many were forced to cover. This buying pressure pushed the stock even higher, creating a feedback loop that intensified the rally. According to S3 Partners, CoreWeave’s “squeeze risk” index hit 100 in early May, the highest possible reading, signaling a perfect storm for shorts.
However, with some short positions already flushed out and borrow rates easing slightly (to around 197%), the worst of the squeeze may be behind us. Available shares to borrow are creeping back up (~250,000 by mid-June), and the stock’s rise has started to slow. That said, the short base remains large, meaning any new positive catalyst, like another major deal or an earnings beat, could reignite a second wave of short covering.
Still, it’s important to remember that high short interest reflects real skepticism. Analysts like Felix Wang at Hedgeye have maintained their bearish outlooks, arguing CoreWeave has no credible path to profitability over the next few years and is burning capital at an unsustainable rate. He, and others, remain short despite the pain, suggesting that conviction among bears hasn’t disappeared.
If fundamentals remain strong and execution persists, shorts can remain underwater. However, should growth stall or insider selling increase after the lock-up expires, the short thesis gains renewed traction with it, and consequently the stock's downside pressure.
Growth Catalysts and Upside Drivers
CoreWeave’s bull case rests on its role as a leading “AI hyperscaler”, providing the specialized cloud infrastructure needed to train and deploy advanced AI models. The company’s fundamentals are undeniably impressive in growth terms. In Q1 2025 (its first report as a public company), CoreWeave delivered $982 million in revenue, up 420% year-on-year. Management’s full-year 2025 guidance calls for ~$5 billion in revenue (nearly 6x 2024 levels). Perhaps more striking is the backlog of contracted business: as of Q1, CoreWeave had $25.9 billion in revenue backlog, multi-year commitments from customers that provide clear visibility into future growth.
This backlog, which the company notes consists of long-term agreements with “success-based” capex matching (i.e., CoreWeave builds capacity as customers ramp up usage), gives credence to the idea that current demand is not just speculative, but contractually locked. For instance, the crown jewel is a five-year, $11.9 billion contract with OpenAI to supply dedicated cloud compute for OpenAI’s model training and inference needs. In addition, just last quarter CoreWeave signed an additional agreement worth up to $4 billion with OpenAI (extending through 2029) to further expand capacity. These deals alone solidify CoreWeave’s revenue base for years, assuming OpenAI continues to execute on its AI roadmap.
Another blockbuster catalyst emerged in Q2 2025: Google became a CoreWeave customer. In June, Reuters reported that CoreWeave has been tapped as a provider of GPU compute capacity to Google Cloud, which in turn will resell that capacity to OpenAI under Google’s new partnership with OpenAI. In effect, CoreWeave is helping Google meet OpenAI’s voracious AI compute demand (augmenting Google’s own data centers), a win-win that diversifies CoreWeave’s revenue beyond Microsoft. (Microsoft was CoreWeave’s largest early customer, accounting for ~62% of 2024 revenue, but Microsoft’s needs may evolve as it rethinks its data center strategy with OpenAI.)
Landing Google as a customer not only reduces over-reliance on any single channel, it also lends credibility and could open the door to other hyperscaler partnerships. Industry insiders note that having a deep-pocketed ally like Google improves CoreWeave’s ability to secure financing for its aggressive expansion plans.
Beyond these mega-deals, CoreWeave is rapidly expanding capacity and rolling out cutting-edge hardware, another upside catalyst for growth. In April 2025, CoreWeave became the first cloud provider to launch Nvidia’s new “Grace Blackwell” superchips at scale. These GPUs (part of Nvidia’s next-gen Blackwell architecture) deliver significant performance gains for AI workloads. CoreWeave’s early deployment enabled marquee clients like Cohere, IBM, and Mistral AI to start using Blackwell systems immediately.
The company even set an industry record in the MLPerf AI benchmark tests with a massive 2,500-GPU Blackwell cluster, demonstrating its technological edge. CoreWeave’s Blackwell-based cloud instances can scale to an astonishing 110,000 GPUs connected by Nvidia Quantum-2 networking, positioning it as a go-to platform for any AI lab needing extreme compute scale. This aggressive capacity build-out (with strong support from partner collaboration with Nvidia) will enable CoreWeave to grab new demand as next-generation AI models spread. For instance, should a wave of new AI startups or corporate AI efforts require cloud GPU muscle in 2025–2026, CoreWeave is better placed than even some of its giant cloud rivals to get them online fast with best-in-class hardware.
Finally, CoreWeave’s extensive partnerships create a virtuous loop: Nvidia’s support provides priority access to new chips; OpenAI’s pledges confirm the model and fuel scale; Google’s partnership diversifies and boosts CoreWeave’s reputation among cloud buyers. Even companies like Core Scientific (a former crypto miner pivoting to AI) have signed 12-year deals with CoreWeave to build out data centers to support its growth (e.g., a $4.7 billion investment for facilities in Texas).
Bottom Line
CoreWeave’s explosive rally has been driven by a rare convergence of structural demand, tight float dynamics, and AI infrastructure tailwinds. While the stock now reflects elevated expectations, its $25.9 billion revenue backlog, deep ties with OpenAI and Google, and first-mover deployment of Nvidia’s Blackwell chips position it as a foundational layer of the AI economy.
The upcoming lock-up expiration may introduce short-term volatility, but for investors with a 1–2 year horizon, any pullback could present an entry point into a company scaling faster and deeper than most competitors. Execution, not sentiment, will determine whether CoreWeave’s valuation grows into a new baseline.
No comments:
Post a Comment