GameStop wants to slash eBay marketing budget

Morgan Stanley similarly doubted the potential cost savings. “On the expense side, we also think the potential opportunities would likely be minimal as physical and digital business require different cost bases, as do 3P marketplaces vs. 1P wholesalers. To add another challenge, GameStop has already undergone a series of large cost cuts,” the research note said.

Morgan Stanley analysts expressed skepticism about “how a deal would be financed given the material valuation gap.” If completed, they said it could end up as the largest leveraged buyout ever, “surpassing the recently announced $55 billion Electronic Arts transaction.”

GameStop said that eBay spent $2.4 billion on sales and marketing in fiscal 2025 but added only 1 million net active buyers, increasing the total from 134 million to 135 million. GameStop said it would cut $1.2 billion from eBay’s sales and marketing budget, arguing that the current spending “is not producing more users on a marketplace with near-universal brand recognition.”

GameStop proposed cutting another $300 million from eBay’s product development expenses and $500 million from general and administrative functions. The combined company will consolidate its finance, HR, real estate, legal, IT, and professional services after the merger to save money, GameStop said.

GameStop touted its own financial performance under Cohen, saying that it “moved from a $381 million net loss in fiscal 2021 to $418 million of net income in fiscal 2025.” eBay meanwhile said in its press release today that its “board and leadership team are executing a focused strategy to drive sustainable growth and long-term shareholder value.”

eBay last week reported Q1 2026 revenue of $3.1 billion, up 19 percent year over year. eBay’s GAAP net income was $512 million, up 2 percent.

GameStop, which is on a different fiscal schedule, reported net sales of $1.1 billion in Q4 2026. That was down from $1.28 billion in the prior year’s fourth quarter. GameStop’s Q4 net income also fell year over year, from $131.3 million to $127.9 million.

Online collaboration is an important tool for companies, big and small. Services like Slack and Microsoft Teams are perhaps the most successful, but their scope is limited primarily to real-time chatting between colleagues, regardless of platform. Another solution is Nextcloud Hub, an open-source, self-hosted content collaboration platform.

Admittedly, I knew little about Nextcloud Hub until recently. However, as a Hostinger client, I noticed the solution was easy to install on my website and recognized it could be a perfect solution for some of our readers.

Denmark generates more than 80 per cent of its electricity from renewable sources. Its wind farms, both onshore and offshore, have made the country a global model for clean energy transition. Its grid operator, Energinet, has spent decades building the infrastructure to support a decarbonised power system. In March, Energinet paused all new grid connection agreements. The reason was not a failure of renewable capacity. It was an explosion of demand from AI data centres that the grid was never designed to accommodate. Denmark, the country that solved clean energy, has become the first Nordic nation to confront a question that the rest of Europe will face within months: what happens when the AI industry’s appetite for electricity exceeds the grid that was built for something else entirely.

The pause

Energinet’s temporary moratorium covers all new large-scale grid connection agreements, not only data centres, but data centres are the proximate cause. Approximately 60 gigawatts of projects are waiting for grid connections in Denmark. The country’s peak electricity demand is roughly 7 gigawatts. The queue is nearly nine times the peak load, and a significant portion of it is data centre capacity. Denmark had around 398 megawatts of installed data centre capacity at the start of 2026, with an additional 208 megawatts under construction and projections to reach 1.2 gigawatts by 2030. Hyperscale facilities, the kind operated by Microsoft, Google, and Apple, account for 60 per cent of the country’s current data centre footprint.

Microsoft alone has committed $3 billion to data centre construction in Denmark between 2023 and 2027. Apple operates a data centre in Viborg. Google has expanded its Danish operations. The hyperscalers chose Denmark for the same reasons the country built its renewable grid: stable governance, reliable infrastructure, cool climate that reduces cooling costs, and abundant wind power. The irony is that the success of Denmark’s green energy model is what attracted the data centres, and the data centres are now overwhelming the grid that made Denmark attractive in the first place.

The 💜 of EU tech

Advertisement

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

The numbers

The scale of AI-driven electricity demand has outrun every forecast. The International Energy Agency reported that data centre electricity consumption surged 17 per cent in 2025, with AI-focused facilities growing even faster. Global data centre electricity use is projected to double by 2030, and power consumption from AI-specific data centres is expected to triple. Startups are racing to curb data centre energy consumption, but the efficiency gains from hardware innovation and cooling technology are being overwhelmed by the sheer volume of new capacity coming online.

A single AI inference task can consume up to 1,000 times more electricity than a traditional web search. Training runs for frontier models require hundreds of megawatts sustained over weeks. The hyperscalers’ combined capital expenditure is projected to exceed $690 billion in 2026, a 36 per cent increase over 2025, and the majority of that spending is directed at data centre construction and the power infrastructure to support it. Efforts to reduce AI’s computational footprint through architectural innovation, including brain-inspired approaches that promise orders-of-magnitude efficiency gains, are years from deployment at scale. In the meantime, the industry is building as fast as grid operators will allow, and Denmark has just demonstrated that grid operators have limits.

The Nordic question

Denmark is not alone among the Nordics in facing this pressure, but it is the first to act. Sweden, Finland, and Norway have all attracted significant data centre investment for the same reasons: renewable energy, cool climates, and stable governance. Sweden’s Lulea, home to a major Facebook data centre, and Finland’s Hamina, where Google operates a facility cooled by Baltic Sea water, are established hyperscale locations. But none of these countries have implemented a grid connection pause.

The Danish moratorium is designed to last three months, during which Energinet will assess how to manage the queue and develop new criteria for prioritising grid connection requests from large energy users. Soren Dupont Kristensen, Energinet’s Chief Operating Officer, described the pause as a “window of opportunity” to rethink regulation. Data centre operators globally are accelerating capital deployment, with Australia’s NEXTDC launching a A$2.2 billion plan anchored by a new campus in Western Sydney. The concern among data centre operators in Denmark is that a three-month pause becomes a longer-term regulatory framework that deprioritises their projects in favour of other industrial uses or residential demand growth.

The tension

The structural tension is between two policy objectives that Denmark has pursued simultaneously: building the world’s cleanest electricity grid and attracting the world’s largest technology companies. Both objectives succeeded. The grid is among the cleanest in Europe. The technology companies arrived. But the grid was designed for a decarbonised industrial economy, not for an AI industry that treats electricity as a raw material consumed at petrochemical scale. Data centres are becoming a financial asset class, with Blackstone preparing the first AI-era data centre REIT. The financialisation of data centre capacity means that capital will flow toward the jurisdictions with the fastest grid connections and the most permissive regulatory environments, and away from jurisdictions that pause.

Denmark’s dilemma is that a pause protects the grid but signals to the hyperscalers that their next billion-dollar facility should be built somewhere else. Ireland faced a similar situation in 2021 when EirGrid imposed a moratorium on new data centre connections in the Dublin area, a pause that lasted more than two years and redirected investment toward other European markets. Denmark’s grid operator has been explicit that the current pause is temporary, but the precedent it sets may be more important than its duration. If Denmark establishes that grid capacity for data centres is not guaranteed, the hyperscalers will diversify their Nordic footprint toward Sweden and Norway, where grid operators have not yet imposed similar restrictions.

The reckoning

Europe has been trying to build a tech ecosystem that can compete with the United States and China. Data centres are the physical infrastructure of that competition: every AI model, every cloud service, and every enterprise deployment runs on hardware that sits in a data centre connected to a power grid. Europe’s total data centre investment is projected to reach €176 billion between 2026 and 2031, but the European Commission’s own analysis warns that future capacity growth will be constrained primarily by grid readiness rather than access to capital. Denmark is the first demonstration of that constraint in practice.

The question Denmark’s pause raises is not whether it was necessary. Given a 60-gigawatt connection queue against 7 gigawatts of peak demand, some form of prioritisation was inevitable. The question is what comes after the pause. If Energinet develops a framework that allocates grid capacity based on economic value, energy efficiency, and contribution to the domestic economy, Denmark could emerge with a model that other European grid operators adopt. If the pause simply delays projects without resolving the underlying capacity mismatch, it will have achieved nothing except pushing investment toward competitors. Denmark built the grid the AI industry wanted. The AI industry showed up with an appetite the grid cannot satisfy. What happens next will determine whether Europe’s cleanest energy market can also be its most competitive data centre market, or whether those two ambitions are, at the scale AI demands, incompatible.

NGC 3137 is a spiral galaxy located 53 million light-years away in the constellation Antlia, and the Hubble Space Telescope has captured it in exquisite clarity. It captured images of this galaxy at six different wavelengths, enabling astronomers to combine ultraviolet, visible, and hydrogen emission data into a single complete picture. At first impression, the image is beautiful, with the galaxy dominating the frame at a sharp angle and its loose arms fanning out like feathers in a summer wind. The older stars in the galaxy’s center have a warm golden tone, whilst the hotter, younger groups of stars on the periphery appear pale blue.

You can see thin dust lanes flowing through the galaxy’s disk, as well as these gorgeous pink spots showing where new stars are being formed as we speak. There are thick clusters of dazzling blue stars that have formed amid the pink clouds of incandescent gas, and each cluster represents a new generation of stars that are just beginning to light. Hubble targeted these features as part of a wider study to examine fifty-five neighboring galaxies. The data is being used to determine how star clusters age from when they first begin to shine brightly until they finally become the older populations that make up the majority of a galaxy. The results provide a vivid picture of life in a spiral galaxy similar to our own Milky Way.

LEGO Technic NASA Artemis Space Launch System Rocket Building Toy for Boys & Girls – STEM Learning…

• BUILD AN OFFICIAL NASA ROCKET – Kids prepare to explore outer space with the LEGO Technic NASA Artemis Space Launch System Rocket (42221) building…
• 3-STAGE ROCKET SEPARATION – Young builders can turn the hand crank to watch the rocket separate in 3 distinct stages: solid rocket boosters, core…
• STEM BUILDING TOY FOR KIDS – This educational rocket kit was created in collaboration with NASA and ESA to showcase the authentic system that will…

So, what exactly is NGC 3137? It belongs to a small group of galaxies known as the NGC 3175 group, which is named after the Local Group, of which we are a member. It comprises of two large spiral galaxies, NGC 3137 and its neighbor NGC 3175, as well as hundreds of smaller, microscopic galaxies that are still being tallied. Astronomers are interested in this group because the layout and movement of the galaxies are quite similar to what we see in the Local Group. Every new discovery about how galaxies behave and interact allows us to gain a better understanding of how our own cosmic backyard operates.

NGC 3137 has a supermassive black hole that is estimated to be sixty million times the mass of our own Sun. There are small dust clouds ringing the core location, adding a lot of dimension to the scene. Because NGC 3137 is inclined slightly towards us rather than directly facing us, we obtain a side-on perspective of the galaxy, which highlights the 3D structure of the arms. There are a few background galaxies in the scene, but they’re just small orange dots, and then there are a few foreground stars that have drifted into the shot; they’re like photobombs, coming out of nowhere and taking the spotlight.

Good morning! Let’s play Connections, the NYT’s clever word game that challenges you to group answers in various categories. It can be tough, so read on if you need Connections hints.

What should you do once you’ve finished? Why, play some more word games of course. I’ve also got daily Strands hints and answers and Quordle hints and answers articles if you need help for those too, while Marc’s Wordle today page covers the original viral word game.

A malicious version of the PyTorch Lightning package published on the Python Package Index (PyPI) delivers a credential-stealing payload targeting browsers, environment files, and cloud services.

The developer disclosed the supply-chain attack on April 30, saying that version 2.6.3 of the package included a hidden execution chain that downloads and executes a JavaScript payload.

PyTorch Lightning is a deep learning framework used for pretraining and fine-tuning AI models. It is a popular package, amassing more than 11 million downloads last month.

The security advisory from the maintainer notes that the malicious execution chain triggers automatically on import and silently spawns a background process.

That process downloads a JavaScript runtime (‘Bun v1.3.13’) from GitHub, and executes a 11.4 MB heavily obfuscated JavaScript payload (‘router_runtime.js’).

In a post over the weekend, Microsoft Threat Intelligence says that Defender detected and prevented the malicious routine on customer environments, and notified the package maintainer.

The payload, which Defender detects as “ShaiWorm,” is an information-stealing malware that targets .env files, API keys, secrets, GitHub tokens, and data stored in Chrome, Firefox, and Brave browsers.

It also interacts with cloud service APIs (AWS, Azure, GCP) to steal credentials and supports arbitrary system command execution.

“lightning==2.6.3 (published on PyPI as py3-none-any wheel) contains a hidden execution chain that silently downloads a JavaScript runtime (Bun) and executes an 11.4 MB heavily obfuscated JavaScript payload upon import lightning,” Lightning AI says in the security advisory.

“This payload contains credential-stealing functionality targeting cloud providers, browsers, and environment files.”

According to Microsoft’s telemetry, the malicious activity affected “a small number of devices” and appears to have been “contained to a narrow set of environments.”

Lightning AI warns that users who ran ‘import lightning’ with version 2.6.3 may have had their secrets, keys, and tokens compromised. In this case, an immediate rotation of all secrets is strongly recommended.

Currently, PyTorch Lightning has been reverted to 2.6.1 on PyPi, which is safe to use.

At this time, it is unclear exactly how the supply-chain compromise occurred, and the package’s publishers are currently investigating how the build/release pipeline was breached.

Additionally, all other recent releases will be audited for similar payloads, and users will be notified via all available channels.

AI chained four zero-days into one exploit that bypassed both renderer and OS sandboxes. A wave of new exploits is coming.

At the Autonomous Validation Summit (May 12 & 14), see how autonomous, context-rich validation finds what’s exploitable, proves controls hold, and closes the remediation loop.

Claim Your Spot

Pinterest reported its first billion-dollar quarter last week. Revenue hit $1.008 billion in the first three months of 2026, up 18 per cent year on year, with monthly active users reaching 631 million for the tenth consecutive quarter of double-digit user growth.

The stock jumped on guidance that projects second-quarter revenue of $1.133 billion to $1.153 billion, a further 14 to 16 per cent increase. Wall Street treated the results as confirmation that Pinterest has finally become the advertising platform it always promised to be. But the interesting part of the earnings is not that Pinterest grew. It is why.

Pinterest did not cross a billion dollars in quarterly revenue by becoming a better social media platform. It crossed a billion dollars by becoming a search engine that happens to show pictures.

The engine

Pinterest processes more than 80 billion searches per month. That number is not a vanity metric. It is the foundation of an advertising model that works differently from every other social platform. When someone opens Instagram or TikTok, they are browsing. When someone opens Pinterest, they are looking for something: a kitchen renovation, a wedding dress, a pair of boots, a recipe for Thursday dinner.

The distinction matters because advertising attached to intent converts at rates that advertising attached to browsing cannot match. Pinterest’s Performance+ suite, its AI-powered campaign automation tool, delivered 24 per cent higher conversion lift in advertiser tests and won 80 per cent of A/B comparisons against manual campaigns. Its return-on-ad-spend bidding system now accounts for 22 per cent of lower-funnel retail revenue on the platform.

These are not engagement metrics. They are commerce metrics, and they explain why advertisers are spending more on Pinterest while scrutinising every dollar they put into platforms that sell attention rather than action.

The AI layer is what changed. Pinterest has spent the past two years rebuilding its advertising stack around machine learning models that match advertiser objectives to user intent signals extracted from visual searches.

When a user photographs a lamp and searches for similar products, or saves a series of pins showing mid-century modern furniture, Pinterest’s models construct an intent profile that is qualitatively different from the interest graphs that Facebook and Instagram build from likes and follows.

The arrival of advertising inside AI platforms like ChatGPT has reframed the conversation about where ad dollars flow, but Pinterest’s results suggest that the most valuable advertising real estate is not inside a chatbot or alongside a social feed. It is at the moment someone is actively searching for something they intend to buy.

The geography

The revenue breakdown tells a story about where the growth is coming from and where it is going. United States and Canada revenue grew 13 per cent. Europe grew 27 per cent. Rest of World revenue surged 59 per cent.

The US and Canada business is mature, generating the majority of Pinterest’s revenue from a user base that has been on the platform for years. The international growth is the early stage of a monetisation curve that Pinterest’s management believes will follow the same trajectory: users arrive for visual discovery, build intent-rich search histories, and become increasingly valuable to advertisers as Pinterest’s AI models learn to match their searches to commercial outcomes.

The 59 per cent Rest of World growth is particularly notable because it is happening on a platform that does not rely on the creator economy, influencer partnerships, or viral content loops that drive growth on TikTok and Instagram.

Pinterest’s international expansion is powered by the same behaviour that drives its domestic business: people searching for things they want. The cultural specificity of those searches, whether it is wedding fashion in India, home décor in Brazil, or street style in Japan, provides the kind of intent data that advertisers in those markets have not had access to on any other platform at this scale.

The contrast

Pinterest’s brand safety advantage has become a competitive moat. Meta faces lawsuits alleging that its platforms have profited from billions of dollars in fraudulent advertising, with internal documents suggesting that a significant share of ad revenue comes from scam accounts. TikTok’s advertising model depends on algorithmic content distribution that periodically surfaces material advertisers do not want their brands associated with. X’s advertising business has contracted since Elon Musk’s acquisition.

Pinterest, by contrast, operates a platform where the content is overwhelmingly aspirational, commercial, and brand-safe by design. People pin products they want to buy, rooms they want to build, meals they want to cook. The content moderation challenge on Pinterest is trivial compared to platforms built around user-generated video and text, and that structural advantage translates directly into advertiser willingness to spend.

OpenAI has shifted ChatGPT’s advertising from impression-based pricing to cost-per-click after its initial $60 CPM launch pricing eroded within weeks, a sign that even the most hyped new advertising platform struggles to prove that its ads drive purchases rather than just visibility. Pinterest does not have that problem. Its advertising model was built from the ground up around commercial intent, and the Performance+ results demonstrate that the AI layer is making the match between intent and advertiser spend more efficient, not less.

The question for advertisers is not whether Pinterest’s ads work. It is whether Pinterest can scale the model fast enough to capture a meaningful share of the $295 billion US digital advertising market before the AI commerce platforms catch up.

The investors

Elliott Investment Management’s $1 billion convertible note investment in Pinterest, disclosed in March, was the activist firm’s bet that the visual search commerce thesis would translate into sustained revenue growth. The Q1 results validated that bet. Pinterest also announced approximately $2 billion in share repurchases, a capital return programme that signals management confidence in the durability of the revenue trajectory.

Elliott’s involvement typically accelerates operational discipline: cost management, margin expansion, and strategic focus on the highest-return initiatives. For Pinterest, that means doubling down on the AI-powered advertising infrastructure that produced the Q1 results rather than diversifying into social features, creator tools, or other distractions that would make Pinterest look more like the platforms it is outperforming precisely because it is not like them.

Alphabet’s Q1 earnings pushed its market capitalisation toward $5 trillion, driven by search advertising revenue that remains the single most profitable business model in technology. Pinterest’s billion-dollar quarter is a fraction of Google’s scale, but the underlying logic is the same: advertising attached to search intent is more valuable than advertising attached to content consumption. Google proved that model with text.

Pinterest is proving it with images. The difference is that Pinterest’s visual search captures intent that text queries often cannot express, the shape of a chair, the colour of a wall, the silhouette of a dress, and converts that visual intent into advertising revenue through an AI pipeline that improves with every search.

The question

Pinterest’s risk is that the same AI capabilities powering its advertising engine are being deployed by competitors who have more users, more data, and more capital. Google’s AI Mode shopping experience combines Gemini with its Shopping Graph to handle product discovery, fit uncertainty, and price timing. Amazon’s Rufus AI assistant now includes an auto-buy function. OpenAI is building conversational ad formats inside ChatGPT that turn advertisements into interactive dialogues. Shopify has launched Agentic Storefronts that make merchant catalogues available inside AI platforms. The commerce layer of AI is being built by companies with resources Pinterest cannot match, and the question is whether Pinterest’s head start in visual search intent data is a durable advantage or a temporary one.

The answer depends on whether visual search remains a distinct category or gets absorbed into the broader AI commerce infrastructure. If consumers continue to use Pinterest as the place where they search for things they want to buy by looking at pictures of things they like, then Pinterest’s intent data is irreplaceable, because no other platform has 80 billion monthly visual searches generating the same density of commercial signals. If AI assistants from Google, Amazon, and OpenAI learn to interpret visual intent as well as Pinterest does, the moat narrows.

Pinterest crossed a billion dollars in quarterly revenue not by winning the social media competition but by refusing to play it. The company built a search engine for visual intent, wrapped it in an AI-powered advertising model that converts searches into sales, and proved that the model scales across geographies and advertiser categories.

Whether that is the beginning of Pinterest becoming a major advertising platform or the high-water mark before AI commerce platforms subsume the category depends on a question Pinterest cannot answer alone: in a world where every platform is adding AI-powered shopping, does the platform that understood visual search first retain the advantage, or does the advantage migrate to whoever has the most compute, the most merchants, and the most aggressive AI deployment?

Pinterest’s billion-dollar quarter is the best argument the company has ever made that the answer is the former. The next four quarters will determine whether the market agrees.