Amazon Falls 1.5% as AWS Says Bahrain Region Cannot Be Restored

Nvidia-backed cloud computing firm Lambda is raising as much as $4 billion in the final round of fundraising before its planned initial public offering, according to a Wall Street Journal report citing people familiar with the matter. The round values the company at $14.5 billion excluding the amount raised and is led by Blackstone and Coatue Management. Lambda is targeting an IPO in 2027, according to a letter sent to the company's limited partners that was viewed by the Journal, and its backlog rose to $50 billion in September from $15 billion in June. A Lambda spokesman declined to comment to the Journal. Lambda, which calls itself a superintelligence cloud company, develops AI cloud infrastructure, was founded in 2012, and is based in California, competing against other neoclouds such as Nebius, Nscale, CoreWeave, and IREN.

IREN Limited reported its fiscal 2026 results and is repurposing its sizable Bitcoin-focused data center capacity toward AI compute workloads, underpinned by new multi-year AI Cloud agreements and ongoing contract expansions with major customers including Microsoft. Among the recent announcements, the five year, US$9,700 million Microsoft AI cloud services contract and the successful delivery of the first 50MW "Horizon 1" deployment are most relevant, anchoring IREN's contracted AI Cloud ARR targets and showing that at least part of its large power and data center footprint is now earning recurring revenue. IREN's narrative projects $13.6 billion revenue and $1.6 billion earnings by 2029, requiring 168.0% yearly revenue growth and about a $2.3 billion earnings increase from -$702.6 million today, and yields a $79.03 fair value, a 95% upside to its current price. Before this news, the most optimistic analysts were assuming revenue could reach about US$25.6 billion by 2029. The pivot carries the risk that heavy capex, debt-funded GPU build outs and still-high crypto exposure could strain cash flow if AI demand or contract renewals underperform.

Boost Run, Inc. has signed a five-year AI cloud services agreement with a leading sovereign AI company carrying a total contract value of $525.6 million, with acceptance of the initial infrastructure expected to begin in early Q2 2027. Under the deal, Boost Run will deliver compute capacity powered by NVIDIA GB300 NVL72 GPUs, along with network storage and CPU node services. The contract brings Boost Run's total contract value to over $2.6 billion, a figure representing total revenue customers have committed to pay over the life of all agreements where the Company's signed contracts have upfront payments. Founder and Chief Executive Officer Andrew Karos said the agreement marks an important step in efficiently deploying the company's financing into contracted revenue and demonstrates the strength and diversification of both its customer base and its GPU fleet. Boost Run said it plans to continue converting its strong pipeline into signed contracts, deploying capacity against contracted demand as the AI compute market continues to expand.

CNBC reports that Microsoft CEO Satya Nadella is facing a fresh challenge in transforming the company for the AI era, after having previously led Microsoft from a software company seen as outdated to a cloud powerhouse now valued at nearly 4 trillion dollars. Although Microsoft was among the first companies to make a big bet on AI through its investment in OpenAI, so far it has no AI product that has clearly emerged as a market leader. Meanwhile, the AI chip market is dominated by Nvidia, and Anthropic and OpenAI have leading models, while Meta is pushing ahead in the Personal AI Agent market and Google's Gemini has gained popularity among both businesses and consumers. Copilot, the heart of its AI strategy, has yet to gain widespread adoption in organizations, and Microsoft shares from the start of the year through October 2 rose only 7%, trailing several large tech stocks and the Nasdaq index, and it has not managed to beat the S&P 500 on a full-year basis since 2023. Nadella is therefore pushing a shift in the business model from per-user monthly subscriptions to consumption-based pricing, with tokens serving as a kind of currency of the AI economy. GitHub Copilot's users grew from 26 million in October last year to 50 million in July, or about one in five GitHub users, and sales accelerated 60% from the previous quarter. Microsoft 365 has a commercial user base of more than 450 million, Copilot charges an additional 30 dollars per user per month, and there are about 30 million Copilot seats. Jared Spataro, an executive in marketing for AI for Work, acknowledged that the company has begun to realize it cannot deliver the best AI results under a flat per-user monthly fee forever, so it is developing a usage-based billing system, or UBB, with its first product being Copilot Cowork, which after launch has seen thousands of customers begin paying and using it. At the same time, the relationship with OpenAI is no longer as close as in the early period, because Microsoft lost its status as an exclusive cloud provider and its rights to use OpenAI's intellectual property changed to non-exclusive. Nadella therefore brought in Mustafa Suleyman, a co-founder of DeepMind, to take charge of AI work starting in 2024, focusing on building a frontier ecosystem rather than having only a frontier model, and having Copilot connect to models from multiple vendors, including Anthropic, OpenAI and Microsoft's own models. This comes amid intensifying competition as Anthropic and OpenAI begin expanding from selling models into the enterprise software market, from coding and cybersecurity to customer service, areas where Microsoft already holds a customer base.