Global Trends
Global Trends
Global market trend intelligence tracks cloud service, regional ISP, national telecom, data centre, and institutional infrastructure signals across global internet infrastructure markets. BTW.
Latest Coverage
Latest from Global Trends
7,509 articles
Global Cloud Services Trends
AWS is growing faster just as Amazon’s AI build turns free cash flow negative
Amazon’s cloud business accelerated in the second quarter, while the group’s trailing free cash flow moved below zero. Those are not opposing stories. They are two stages of the same investment cycle. AWS sales rose 37% to $42.2 billion and segment operating income reached $16.6…
Global Datacenter Trends
An Ohio automotive factory is being reassigned to the AI cooling cycle
Mitsubishi Electric’s planned data-centre cooling plant in Mason, Ohio, begins with an asset from another industrial era: part of an existing automotive-electrical production site. The company will establish MEHITS US in August, invest approximately $30 million in renovation and…
Global Datacenter Trends
AEP is reserving turbines for a 69GW load book that has not reached the meter
American Electric Power has begun matching contracts for future electricity demand with reservations for future generation equipment. It added 6GW of signed load agreements in the second quarter, primarily in Texas, taking contracted load growth through 2030 to 69GW. The customer…
Global Datacenter Trends
Exelon has put a price on the difference between a data-centre request and a credible queue
Exelon now describes about 11GW of prospective large load as “high probability”, but the useful disclosure is not the adjective or even the scale. It is the evidence underneath it. Projects in that group have reached advanced design or obtained FERC-approved Transmission Security…
Global Cloud Services Trends
Meta’s invisible infrastructure queue has reached $279bn
Meta spent $31.08 billion on capital equipment and finance-lease principal in the second quarter. The larger number is waiting outside the period’s visible expenditure: $278.99 billion of operating and finance lease obligations for data centres, colocation and certain network…
Global Datacenter Trends
Europe’s AI gigafactory tender is a contract for demand
The European Union has opened procurement for up to seven AI gigafactories, backed by as much as €10 billion of EU and national money and an expectation of at least €20 billion in private investment. The significant instrument is not the subsidy alone. EuroHPC and 18 Member…
Global Datacenter Trends
Microsoft’s 88 data centres are a throughput claim, not a map
Microsoft says it brought 31 new data centres online across five continents in its final fiscal quarter and 88 across the year. The larger revelation is not the pin count. In the same quarter it added another gigawatt of capacity, while cutting the time needed to move new GPUs…
Global Datacenter Trends
A management port is not a breach—but 24,650 pre-login replies redraw the risk
Lava’s internet measurement does not show 24,650 compromised servers. It shows something more precise: 24,650 public IPMI endpoints returned password-derived authentication material before a client had authenticated. That distinction separates evidence from alarm. It also exposes…
Global Cloud Services
EU fines Google €890m over Search and Play
Google has 60 days to change Search presentation and Play steering rules after two DMA fines totalling €890 million.
Global Cloud Services Trends
OpenAI’s `invalid_prompt` incident made a provider fault look like a caller mistake
OpenAI recorded elevated API errors under the label `invalid_prompt` for one hour, twenty-four minutes and fifty-two seconds on 29 July. That name usually sends an engineer toward the request body. The provider’s incident record did not disclose the accompanying HTTP status…
Global Cloud Services Trends
GitHub’s outage numbers trace six products back to one lost network quarter
GitHub’s July 29 root-cause update turns a broad service incident into a physical capacity story. On July 24, links between one compute cage’s spine switches and an availability zone’s aggregation layer lost connectivity; the surviving paths saturated, one quarter of interconnect…
Global Cloud Services Trends
Newforma’s seven-year AWS commitment turns a cloud roadmap into a dependency timetable
Newforma and Amazon Web Services already had a working relationship. The July 28 announcement matters because it gives that relationship a seven-year horizon and names the work inside it: modernization of Project Center and Konekt, production AI workloads on Bedrock, and a…
Global Cloud Services Trends
Cloudflare fixed its Tunnel service before every customer had necessarily recovered
Cloudflare’s July 28 incident ended at the provider, but its final guidance left work at the customer edge: anyone still unable to reach private resources was told to restart the affected `cloudflared` connectors. That distinction—service restored versus access restored—is the…
Global National Telecom Trends
SKYWAVE and Iridium are trying to make the network disappear from the machine operator's decision
A mining truck should report a fault because it has a working path, not because its operator knows whether satellite, cellular or Wi-Fi is available. SKYWAVE and Iridium have begun integrating Iridium Short Burst Data into SKYWAVE terminals for heavy-equipment manufacturers. The…
Global National Telecom Trends
Amazon Leo’s 5,105-satellite filing buys an option, not a mobile service
Amazon Leo has asked the US Federal Communications Commission for authority to deploy up to 5,105 direct-to-device satellites, with deployment proposed to begin in 2028. The filing defines a network architecture and a route to mobile-operator partnerships; it does not yet…
Global Cloud Services Trends
DigitalOcean identified Agent Runtime HTTP 500 errors but had not resolved them
DigitalOcean began investigating HTTP 500 responses from Agent Platform requests at 18:31 UTC and said at 19:44 UTC that the issue was identified and a fix was being implemented. Agentic Inference Cloud Agent Runtime was affected; no resolution was posted by 03:43 UTC on 28 July.
Global Cloud Services Trends
GitHub resolved a 16-minute GraphQL incident before explaining its scope
GitHub reported degraded API-request performance from 03:53:19 UTC, posted mitigation at 04:09:01 and resolution at 04:09:10—about 16 minutes. The incident title named GraphQL API requests while the update text referred to API Requests more broadly; magnitude and cause were not…
Global Cloud Services Trends
GitHub’s 105-minute incident turned one latency source into four work queues
A major GitHub incident affected Actions, Issues, Webhooks and Pull Requests for 1 hour 45 minutes 20 seconds. Slow job starts, stale search and backlogs did not merely delay individual clicks: they separated code, automation, event delivery and review into queues that teams had…
Global Cloud Services Trends
IBM is buying control of a quantum research stack without revealing its price
IBM has signed a definitive agreement to acquire HRL Laboratories from Boeing and General Motors. The deal brings silicon-spin qubits, sensing, materials, cryogenics, controls, interconnects and packaging under one buyer, but the financial terms are undisclosed and the…
Global Cloud Services Trends
OpenAI’s cross-product errors exposed a reliability boundary below its own services
OpenAI reported elevated errors across listed API, ChatGPT and Codex components and said it was implementing mitigation with a downstream infrastructure provider. At the fixed 17:23:49.505 UTC reporting cutoff, the incident remained identified and unresolved, while the vendor…