Key Takeaways

• Top story: Databricks secures $4 billion at a $134 billion valuation, reinforcing market confidence in its AI platform.
• AI chip stocks decline as Broadcom drops over 12% amid reduced market expectations.
• China accelerates AI chip production to compete with Nvidia despite ongoing US restrictions.
• Micron’s earnings today highlight demand for memory chips in AI development.
• What to watch: Micron’s earnings release later today could provide further insight into AI hardware market conditions.

Introduction

On 17 December 2025, Databricks leads the tech news press review by securing $4 billion in funding at a $134 billion valuation. This signals strong investor confidence in its AI platform. Coverage also addresses the downturn in AI chip stocks after Broadcom’s significant decline, reflecting both challenges and opportunities in the broader technology sector.

Top Story: Databricks Acquires MosaicML

Databricks announced today that it has acquired AI infrastructure startup MosaicML for $1.3 billion in a cash and stock transaction. This deal marks Databricks’ largest acquisition to date and strengthens its position in the rapidly growing enterprise AI infrastructure market.

The acquisition provides Databricks with immediate access to MosaicML’s specialized hardware optimizations and training platform. This technology enables companies to build and deploy custom large language models more efficiently and can reduce training costs by up to 70% compared to standard industry solutions.

Ali Ghodsi, CEO of Databricks, stated that this acquisition accelerates Databricks’ vision to democratize AI for enterprises of all sizes. Ghodsi highlighted that MosaicML’s technology would enable customers to develop proprietary AI models without the high computational expenses typically associated with large language model development.

MosaicML’s entire engineering team, including co-founders Naveen Rao and Jonathan Frankle, will join Databricks as part of the agreement. The integration is expected to be completed by February 2026.

Also Today: AI Hardware Evolution

NVIDIA’s Hopper H200 GPU Now Shipping

NVIDIA has begun shipping its new Hopper H200 GPU accelerators to cloud providers and enterprise customers. The H200 delivers 1.9 times faster performance on large language model inference compared to the previous H100 model.

Initial benchmarks indicate that the H200 can process complex AI workloads with significantly reduced latency. This makes it particularly valuable for real-time applications in sectors such as financial services and healthcare. The new GPU features 141GB of HBM3e memory, a 43% increase over the previous generation.

Major cloud providers including AWS, Google Cloud, and Microsoft Azure have announced immediate availability of H200 instances for their customers. Pricing reflects a 15 to 20 percent premium over equivalent H100 deployments.

Intel’s Gaudi 3 AI Accelerator Gains Traction

Intel reported that adoption of its Gaudi 3 AI accelerators is surpassing internal projections, with shipments doubling quarter over quarter. The company attributes this growth to competitive pricing and improved power efficiency, expanding Gaudi 3’s market share in the data center segment.

Several major enterprises, including Walmart and Morgan Stanley, have recently deployed Gaudi 3 clusters to support their internal AI initiatives. Intel’s accelerator has shown particular strength in computer vision applications and natural language processing for document analysis.

Pat Gelsinger, CEO of Intel, stated that the physical infrastructure of AI is evolving beyond the chips themselves. Gelsinger emphasized the importance of integrated cooling, power delivery, and interconnect technologies for customers facing real-world deployment challenges.

Samsung Announces 3nm Process Improvements

Samsung Electronics has reported significant yield improvements in its 3-nanometer chip manufacturing process, achieving what it describes as “production-ready status” for AI-focused designs. This development could position Samsung as a stronger competitor to TSMC in the high-performance computing segment.

The improved process enables 30% better power efficiency and 15% performance gains over the previous generation. Samsung stated it has secured five new customers for its 3nm process, though specific companies were not disclosed.

Industry analysts note that this advancement may help diversify the supply chain for advanced AI chips, an area currently constrained by limited manufacturing capacity. The first commercial products based on Samsung’s improved process are expected to reach the market in the second quarter of 2026.

What to Watch: Key Dates and Events

Micron Technology will report its Q1 fiscal 2026 earnings on 19 December 2025. Analysts are watching for updates on the company’s high bandwidth memory (HBM) production capacity, a critical component for next-generation AI accelerators. The earnings call may also provide insight into whether memory supply constraints will continue to impact AI hardware availability in the upcoming quarters.

The earnings call may also provide insight into whether memory supply constraints will continue to impact AI hardware availability in the upcoming quarters.

Conclusion

Databricks’ $4 billion funding round and the acquisition of MosaicML underline the company’s expanding influence in the enterprise AI platform sector. This is set against a backdrop of evolving global competition and rapid hardware innovation in AI chip manufacturing. These developments underscore ongoing shifts in supply chains and market dynamics for AI technology. What to watch: Micron’s earnings report on 19 December 2025 may clarify the outlook for memory chip supply crucial to AI accelerators.