Welcome to The Median, DataCamp’s newsletter for February 27, 2026.
In this edition: Nvidia’s record Q4, Anthropic’s standoff with the Pentagon, OpenAI’s massive investment round, Google’s new Nano Banana 2, AI distillation attacks, and our 2026 Data & AI Literacy Report.
This Week in 60 Seconds
Nvidia Reports Stellar Q4 Revenue Amid Cautious Market Reaction
Nvidia announced record-breaking financial results for the fourth quarter of fiscal 2026, posting $68.1 billion in revenue, which represents a 73% increase from the previous year. This growth was heavily driven by its Data Center segment, which generated $62.3 billion as enterprise adoption of AI computing infrastructure continues to expand. Despite beating estimates and projecting $78.0 billion in revenue for the first quarter of fiscal 2027, the company’s shares fell by over 5% following the announcement. The broader tech market also experienced a downturn, with the Nasdaq dropping 1.2%, as investors remain cautious about the long-term sustainability of AI capital expenditures.
Anthropic Refuses Pentagon Demands for Unrestricted AI Use
Anthropic is engaged in a high-stakes standoff with the Department of War after refusing to remove ethical safeguards from its military contracts. Secretary of War Pete Hegseth issued an ultimatum demanding that the company permit its Claude models to be used for any lawful military purpose. However, CEO Dario Amodei publicly rejected the demands, stating the company will not allow its AI to be used for mass domestic surveillance or fully autonomous weapons, citing concerns over technical reliability and democratic values. In case of a refusal, the Pentagon had threatened to label Anthropic a supply-chain risk or invoke the Defense Production Act to force compliance by February 27.
OpenAI Annouces A Massive $110B Investment Round
OpenAI has announced a new $110 billion investment round at a $730 billion pre-money valuation to help support its growing infrastructure and global computing needs. The funding includes $50 billion from Amazon, $30 billion from SoftBank, and $30 billion from NVIDIA, alongside strategic partnerships that secure next-generation compute capacity, including extensive use of NVIDIA’s Vera Rubin systems for training and inference. These expanded resources aim to maintain service reliability and access for a user base that has now reached 900 million weekly active ChatGPT users, 50 million consumer subscribers, and 9 million paying business customers.
Google Launches Nano Banana 2
Google has introduced Nano Banana 2, also known as Gemini 3.1 Flash Image, a new model designed to combine the visual quality of its Pro predecessor with significantly faster processing speeds. The update integrates real-time web search data to improve factual accuracy and features enhanced text rendering capabilities, including text translation directly within images. For complex workflows, the model supports subject consistency for up to five characters and 14 objects, while allowing users to generate assets in various aspect ratios up to 4K resolution. Nano Banana 2 is now the default image generation model across multiple Google platforms, including the Gemini app.
Anthropic Reports Distillation Attacks By DeepSeek, Moonshot, and MiniMax
Anthropic has uncovered coordinated distillation campaigns by three AI laboratories (DeepSeek, Moonshot, and MiniMax) designed to illicitly extract capabilities from its Claude models. By deploying approximately 24,000 fraudulent accounts through commercial proxy services, these organizations generated over 16 million exchanges to train their own systems on Claude’s advanced reasoning, coding, and tool-use outputs. Anthropic stated that this practice allows competitors to acquire powerful AI capabilities at a fraction of the traditional cost and time, while bypassing the critical safety guardrails built into US-based models.
DataCamp Releases the 2026 State of Data and AI Literacy Report
This week, our team at DataCamp published the 2026 State of Data & AI Literacy Report to map the current skills landscape. Built on independent research conducted by YouGov between December 2025 and February 2026, the study surveyed 517 leaders from organizations with 500 or more employees across the United States and the United Kingdom. The findings reveal a noticeable paradox in the modern workplace: while 88% of leaders consider basic data literacy important for day-to-day work and 72% say the same for AI literacy, roughly 60% still observe significant capability gaps within their organizations. We unpack this report in our Deeper Look section below.
New: The 2026 State of Data and AI Literacy Report
A Deeper Look at This Week’s News: The State of Data & AI Literacy in 2026
Our team at DataCamp has released the 2026 State of Data & AI Literacy Report, built on independent research surveying over 500 enterprise leaders across the US and UK to map the current skills landscape.
The data and AI skills paradox
The findings reveal a distinct contradiction in the modern enterprise: expectations have reached a critical threshold, with 88% of surveyed leaders now considering basic data literacy essential for day-to-day work, and 72% holding the exact same standard for AI literacy.
Yet despite this high demand, roughly 60% of these leaders report significant data and AI skill gaps in their current workforce.
How this paradox might actually be an opportunity
When companies pair their AI investments with structured, organization-wide upskilling programs, they are twice as likely to report significant positive ROI from their AI tools (42% compared to the 21% baseline).
This organizational return is driven directly by individual output. According to the survey, 76% of leaders observe that data-literate employees consistently outperform their less-equipped peers.
The top benefits of AI and data literacy
When employees understand how to use data and AI effectively, the organizational benefits extend beyond basic task efficiency. According to the report, 54% of leaders identify faster decision-making as a primary benefit of data literacy, with 48% observing the same acceleration from AI skills.
Accuracy and creativity also see significant improvements across the board. Nearly half of the surveyed executives directly link these capabilities to a stronger ability to innovate and more accurate strategic choices.
How to stand out in the 2026 job market
For the individual employee, the shift toward baseline AI expectations presents a significant financial opportunity. According to the report, strong data and AI skills command a consistent salary premium, typically ranging between 10% and 30%.
However, because basic proficiency is increasingly treated as a standard requirement, candidates must go beyond simple software operation to differentiate themselves. Our report identifies four specific ways professionals can rise above the average:
1. Combine technical with soft data and AI skills
Leaders place the highest value on professionals who pair technical abilities with the capacity to interpret information and apply it in context. Strong candidates translate AI outputs into actionable narratives that non-technical stakeholders can understand and trust.
2. Prioritize adaptability and critical thinking
As AI tools become more accessible, the primary workplace risk shifts from a lack of access to a lack of judgment. Employers actively search for individuals who evaluate and question the outputs they rely on rather than accepting them at face value. Adapting quickly to new models while maintaining a healthy skepticism is a highly sought-after trait.
3. Engage with data quality, governance, and trust issues
Real-world data is rarely perfect. Employers consistently point to inconsistent definitions, unclear ownership, and weak governance as major barriers to effective decision-making. You can stand out by showing you know how to assess data quality, ask the right questions about where information originates, and work safely within privacy and ethical constraints.
4. Demonstrate impact, not just skill
Ultimately, organizations do not reward skills in isolation. They reward better outcomes. The most competitive applicants frame their experience by explaining how applying a specific tool allowed them to accelerate a process, improve a decision, or drive tangible productivity gains for their team.
To explore the complete findings and discover practical steps for closing the skills gap, read our full 2026 State of Data & AI Literacy Report.
Industry Use Cases
Samsung Focuses on Proactive and Agentic AI in Galaxy S26 Series
Samsung has introduced the Galaxy S26 series with a focus on proactive AI features designed to anticipate user needs and automate routine tasks. The devices utilize contextual tools to suggest relevant photos for sharing, provide timely event reminders, and check calendar conflicts, minimizing the need to switch between applications. External AI agents from Gemini and Perplexity allow users to execute multi-step processes, such as booking a ride, using simple natural language prompts. The system also incorporates AI-powered call screening to identify unknown callers and machine learning-driven privacy alerts that monitor app permissions in real time. Read more in this article from Samsung.
Eli Lilly Launches Pharmaceutical AI Supercomputer For Drug Discovery
Eli Lilly has activated LillyPod, an AI supercomputer equipped with over 1,000 NVIDIA Blackwell Ultra GPUs designed to accelerate pharmaceutical research. Capable of delivering more than 9,000 petaflops of computing performance, the system allows scientists to bypass the physical constraints of traditional laboratories. Instead of being limited to testing a few thousand molecular concepts annually, researchers can now simulate and evaluate billions of molecular hypotheses computationally before conducting physical experiments. Read more in this article from the NVIDIA blog.
Oura Launches Proprietary AI Model For Women’s Health
Smart ring manufacturer Oura has introduced its first proprietary AI model to power its Oura Advisor chatbot, providing personalized insights across the entire reproductive health spectrum from early menstrual cycles to menopause. Developed alongside board-certified clinicians, the system combines established medical research with a user’s specific biometric data, including sleep, activity, stress, and cycle history, to offer personalized health information. The tool is designed to be emotionally supportive and reassuring, though it is not intended to replace a doctor for medical diagnoses or treatment plans. Read more in this article from TechCrunch.
Tokens of Wisdom
The physical world around us runs on software already anyway. There are sensors everywhere, but nothing is particularly intelligent. It’s just automata.
—Ivan Poupyrev, CEO at Archetype AI
This week on the DataFramed podcast, we sat down with Ivan Poupyrev to examine physical AI. We discussed physical AI beyond robotics, turning IoT sensor streams into insights, why physical foundation models differ from LLMs, and much more.