Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
CoreWeave did not lose Core Scientific because an official filing blamed “AI mania.” The verified immediate cause was that Core Scientific shareholders failed to approve the proposed merger, leading Core Scientific to terminate the agreement on October 30, 2025. On that same day, CoreWeave announced an agreement to acquire Marimo, the open-source company behind a reactive Python notebook.
The two events are linked by timing, not by a formal replacement transaction. Core Scientific represented physical infrastructure—data centers, power, and high-density capacity. Marimo represents the developer workflow that can bring users onto an AI cloud. For investors, the contrast illustrates the risks of stock-funded infrastructure M&A and the appeal of software-led distribution. For developers, it explains why CoreWeave now has a stake in the notebook layer of AI development.
The timeline: two deals converged on October 30
- July 7, 2025: CoreWeave announced its proposed acquisition of Core Scientific, describing the transaction as a way to accelerate vertical integration and strengthen the infrastructure behind its AI cloud. CoreWeave’s announcement framed the deal around data-center capacity, power, and AI and high-performance-computing demand.
- July–October 2025: The deal moved through the shareholder-approval process. The transaction involved CoreWeave equity, making the value of the offer sensitive to CoreWeave’s share price and prospects.
- October 30, 2025: Core Scientific shareholders failed to provide the required approval. Core Scientific then terminated the merger agreement and remained an independent Nasdaq-listed company under ticker CORZ. Its regulatory filing identifies the failed shareholder vote as the reason.
- October 30, 2025: CoreWeave announced its definitive agreement to acquire Marimo. The financial terms were not disclosed. CoreWeave described the deal as a way to unify the generative-AI developer workflow with its cloud platform and existing Weights & Biases tooling.
- June 1, 2026: Marimo said its hosted molab notebook environment was running on CoreWeave Cloud with GPU access in public preview.
The same-day announcements make for a compelling headline, but the evidence does not show that CoreWeave abandoned Core Scientific in order to buy Marimo. They were separate strategic developments aimed at different parts of the AI stack.
What CoreWeave wanted from Core Scientific
Core Scientific is involved in high-density data-center colocation and digital-asset mining. CoreWeave’s proposed acquisition was fundamentally an infrastructure transaction, not a conventional software purchase.
#1 Best Overall
The strategic prize was greater control over:
- Data-center sites and electrical power;
- High-density computing facilities;
- Expansion and construction timelines;
- The conversion of AI demand into deployable GPU capacity; and
- A bottleneck that can constrain AI-cloud growth even when customers are ready to pay.
A secondary technical report estimated that the proposed transaction would add approximately 1.21 gigawatts of gross power capacity, with additional expansion potential. That figure should be treated as an attributed estimate rather than a current, independently verified capacity figure. The important point is strategic: CoreWeave was seeking more ownership and control of the physical layer beneath its GPU cloud.
Why the acquisition failed
The legal answer is simple: the required shareholder approval was not obtained. Core Scientific’s October 30 filing does not identify antitrust opposition, a financing failure, or a formal withdrawal by CoreWeave as the cause of termination.
That does not mean market conditions were irrelevant. It means “AI mania tanks the deal” is an interpretation, not the official explanation.
Free tools Windows power users keep installed
One-click scans. No signup required.
Where AI-market sentiment may have mattered
Stock-funded acquisitions are especially vulnerable to volatility. If the buyer’s share price falls, the stock being offered to the target’s shareholders may look less attractive than it did when the transaction was announced. Shareholders must then weigh the offer against:
- The revised market value of the consideration;
- The risks of owning shares in a highly capital-intensive AI company;
- Whether future AI demand justifies continued data-center expansion;
- Execution risks involving construction, permitting, power, and operations; and
- Potential dilution and exposure to CoreWeave’s valuation.
That is the defensible version of the “AI mania” thesis: a deal negotiated during an AI-infrastructure boom can become harder to approve when valuations and share prices are volatile. It is not proof that changing sentiment alone killed the transaction.
Investors reviewing the economics should consult the Core Scientific merger proxy and CoreWeave’s SEC filing rather than rely on rounded figures from headlines. The exact exchange ratio, implied valuation, and market-price impact require context from those documents.
What Marimo adds
Marimo is an open-source reactive Python notebook for data science, machine learning, AI, and scientific computing. Its central distinction is that notebooks are stored as pure Python rather than notebook-specific JSON.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsAccording to Marimo’s product description, its notebooks are designed to:
- Update dependent cells reactively when code changes;
- Reduce hidden execution state and improve reproducibility;
- Work more naturally with Git, code review, and packaging;
- Run as notebooks, scripts, modules, pipelines, or interactive applications;
- Support SQL, databases, and data lakes; and
- Connect experimentation with deployable Python software.
That does not make Marimo universally better than Jupyter. Jupyter remains the more established and portable ecosystem, particularly for organizations operating JupyterHub, supporting multiple languages, or relying on mature extensions. Marimo’s appeal is narrower and more specific: a reactive, Git-friendly Python workflow that can move from exploration toward an application without treating the notebook as a disposable artifact.
Why CoreWeave wants a notebook
CoreWeave has historically competed for infrastructure demand: GPU compute, storage, networking, and related cloud services. A developer tool gives it a potential entry point earlier in the customer journey.
- A developer experiments with data or a model in a notebook.
- The workload needs CPU, memory, storage, or GPU access.
- The notebook becomes an interactive application, script, or pipeline.
- The developer may then need hosted inference, training, deployment, or experiment tracking.
- CoreWeave can try to capture more of that workflow through its cloud and adjacent Weights & Biases tooling.
This is strategic analysis, not evidence that the combination has already produced a particular revenue increase or customer-conversion rate. The potential benefit is distribution and customer stickiness: Marimo can showcase CoreWeave infrastructure to developers who might otherwise encounter the company only after making a cloud decision.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteThe trade-off is that open-source software is difficult to monetize directly. Developers can run marimo locally, on Jupyter-style infrastructure, or on competing clouds. CoreWeave would need to benefit indirectly through cloud consumption, hosted services, enterprise features, or cross-selling—not simply through ownership of the notebook code.
What molab looked like in 2026
On June 1, 2026, Marimo announced that its hosted molab environment was running on CoreWeave Cloud in public preview. Marimo listed:
- 4 CPUs and 32 GB of RAM by default;
- An optional NVIDIA RTX Pro 6000 Blackwell GPU;
- 96 GB of GPU memory;
- Up to 125 TFLOPS, according to Marimo;
- Sessions lasting up to 12 hours; and
- Free access while usage remained reasonable.
These are public-preview terms, not a permanent or unlimited free-cloud promise. Marimo says resource parameters may change if demand exceeds available capacity.
molab also is not a general-purpose production server. Its restrictions prohibit activities including crypto mining, remote proxies, file hosting, compute resale, and non-interactive jobs. Notebooks are described as public but not discoverable by default, which is not the same as private enterprise storage. Geographic restrictions may also apply.
For low-friction demonstrations, education, prototypes, and interactive AI or data notebooks, molab can be useful. It is a poor fit for confidential workloads, unattended training, persistent services, guaranteed capacity, or formal production SLAs.
Does Marimo remain open source?
Marimo said that after joining CoreWeave its notebook would remain free, open source, and permissively licensed, with the existing open-source roadmap continuing and the team receiving additional resources. That is the company’s stated commitment at the time of the acquisition. It should not be read as a permanent guarantee that every future hosted feature, usage limit, or commercial offering will remain unchanged.
Users can install the open-source notebook locally with:
pip install marimo
marimo tutorial introLocal installation provides more control over data, dependencies, persistence, and hardware, but the user must manage the environment and any GPU infrastructure.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Core Scientific versus Marimo: a strategic comparison
| Issue | Core Scientific | Marimo |
|---|---|---|
| Layer of the stack | Physical infrastructure | Developer software and hosted notebooks |
| Primary asset | Sites, power, facilities, and capacity | Open-source Python workflow and developer adoption |
| Capital profile | Capital-intensive and operationally complex | Software-led, though hosted compute still costs money |
| Strategic objective | Control more AI/HPC capacity | Reach developers and connect experimentation to cloud usage |
| Main risks | Valuation, dilution, construction, permitting, and demand risk | Open-source monetization, user neutrality, compute costs, and competition |
Marimo therefore does not solve the problem Core Scientific was intended to address. It does not add data-center power, eliminate CoreWeave’s capital intensity, guarantee GPU availability, or replace the physical-capacity strategy.
Best Value
What this means for investors
The failed Core Scientific transaction is a reminder that infrastructure M&A can be strategically attractive and still difficult for shareholders to approve. The offer must remain compelling after accounting for stock volatility, dilution, execution risk, and uncertainty about future AI demand.
The Marimo acquisition points to a different strategy: build a software layer that can help CoreWeave reach developers before they commit to infrastructure. That could make the cloud more differentiated than a provider selling GPU capacity alone, but the economic payoff remains an open question.
The most accurate conclusion is not that CoreWeave replaced a $9 billion infrastructure deal with a notebook acquisition. The transactions were not equivalent in size, purpose, or risk. CoreWeave lost a proposed physical-capacity expansion and gained a developer-facing software asset on the same date.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →What this means for developers
Marimo is worth considering when a team values pure-Python notebooks, reactive execution, Git workflows, SQL and data connectivity, and the ability to turn an experiment into a script or interactive application.
Jupyter or JupyterHub may remain the better choice when an organization needs broad language support, extensive existing integrations, self-managed infrastructure, or established multi-user administration. Google Colab may be more convenient for users deeply invested in Google Drive and Google Cloud. A self-hosted marimo deployment offers more control over privacy, authentication, storage, and GPU selection than molab.
For hosted GPU work, compare the complete economics—not just an advertised hourly rate. GPU model, utilization, storage, networking, egress, idle time, quotas, support, and deployment requirements can matter more than the notebook interface. Google’s Colab Enterprise pricing page, for example, lists approximate accelerator rates that vary by region and configuration. CoreWeave’s current pricing should be checked directly with the provider rather than inferred from third-party estimates.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
