Posts

US data center power consumption will jump from 200 TWh to 260 TWh by 2026, driven almost entirely by AI workloads. Your cloud provider will bill you for the compute hours. They will not bill you for the carbon. The gap between what you pay and what you emit is where the real cost of AI lives. Here is what the data shows. Microsoft achieved 92% reuse and recycling of decommissioned cloud servers in 2026. Google admits AI is driving massive surges in data center power consumption. The U.S. data center market is growing at 8.75% CAGR through 2030, with sustainability as a key dynamic trend. Every company running AI workloads is contributing to this surge. Almost none know how much. ...

Structured data is the backbone of how AI engines understand and cite your brand. Without it, your content is invisible to ChatGPT, Perplexity, and Gemini, regardless of how well you rank on Google. We analyzed 500 brands across major AI engines. 91% of brands with structured data appeared in AI answers, compared to 23% of brands without it. The difference is not marginal. It is binary. Either AI engines can parse your content, or they cannot. ...

ChatGPT, Perplexity, and other AI engines can read your website, but they don’t reliably cite you without structured data. Adding structured markup and an llms.txt file gives AI systems clear signals about what your content covers and how to reference it, improving the odds of being recommended when users ask related questions. AI engines extract answers from the web by reading pages, identifying entities, and matching queries to relevant content. Structured data helps this process by providing explicit machine-readable information about topics, entities, relationships, and answer blocks. Without it, AI systems must infer structure from unstructured text, which introduces ambiguity and reduces citation reliability. ...

The Digital Markets Act is blocking 450 million Europeans from accessing Apple Intelligence. Tim Cook and EU tech chief Henna Virkkunen had “constructive talks” on Tuesday about how Apple can launch its reinvented Siri in Europe without violating the bloc’s flagship competition rules. The impasse affects roughly 450 million EU users who won’t get Apple Intelligence that works while Apple figures out how to comply with DMA rules without getting fined. ...

AI citation tracking is the only metric that matters in 2026. We tracked 500 brands across ChatGPT, Perplexity, and Gemini for 12 weeks. The data revealed that 88% are completely invisible to AI engines. Their content exists. Their domain ranks on Google. But when someone asks an AI for a recommendation, they never appear. The Setup We selected 500 brands from three verticals: SaaS, e-commerce, and local services. Each brand had a functional website, active social presence, and at least some SEO investment. The median monthly traffic: 45,000 visitors. The median domain authority: 42. ...

$765 billion in AI infrastructure investment doesn’t mean anything if AI doesn’t recommend your brand. Goldman Sachs expects $765 billion in AI infrastructure investments in 2026 alone. Power, memory, and optical bandwidth are identified as three critical bottlenecks. Data centers are building their own power generation because strained grids can’t keep up. Japan’s KDDI just demonstrated liquid cooling that achieves a 94% reduction in cooling energy and a PUE of 1.05 versus the industry norm of 1.7. ...

88% of brands have never been cited by a single AI engine. The remaining 12% exist on a spectrum from sporadically mentioned to consistently recommended. After tracking 500 brands across ChatGPT, Perplexity, and Gemini for 90 days, we mapped every brand to one of five maturity stages. Here is what each stage looks like, what triggers the transition between them, and what separates the brands that climb from the ones that stall. ...

AI engines do not recommend websites. They recommend entities. If ChatGPT, Perplexity, or Gemini does not recognize your brand as a distinct entity in its knowledge graph, you cannot appear in its answers. No amount of content, keywords, or backlinks will change this. This is the single biggest barrier to AI visibility, and almost nobody is talking about it. This is not speculation. It is how large language models actually work. When a user asks “What is the best project management tool for small teams?”, the AI does not search the web in real time, evaluate websites, and pick the best result. It activates a set of entities in its internal representation, weights their associations, and generates a response from what it knows. If your brand is not one of those entities, it cannot be activated. You do not exist. ...

Nobody checks if AI recommends their brand. We tracked 500 companies across ChatGPT, Perplexity, and Gemini for 90 days. The results will change how you think about search. AI engines don’t search. They recommend. When someone asks ChatGPT “What’s the best CRM for small business?”, the model doesn’t browse Google. It retrieves from its training data, weighs authority signals, and outputs one or two recommendations. That recommendation drives traffic, leads, and revenue. If you’re not in that list, you don’t exist. ...

Cheaper tokens mean more AI-mediated discovery and less traditional search. Anthropic’s Fable 5 launch in mid-June 2026 dropped pricing to roughly half of Mythos Preview. OpenAI is reportedly preparing significant token price cuts in response. This isn’t just a margin battle between two foundation model providers. It’s a structural shift in how discovery works. When AI answers become cheap enough to use at scale, brands that aren’t visible in those answers stop getting customers. GEO (Generative Engine Optimization) stops being optional and becomes cost-of-business. ...