Educational Technology and Change Journal

By Jim Shimabukuro (assisted by Claude)
Editor

1. NVIDIA

NVIDIA Corporation is headquartered in Santa Clara, California, and was founded in 1993 by Jensen Huang, Chris Malachowsky, and Curtis Priem. It is a fabless semiconductor company — meaning it designs its chips but outsources manufacturing, primarily to TSMC in Taiwan. Today, with a market capitalization that has surpassed four trillion dollars, NVIDIA stands as one of the most valuable companies in the history of global business.

NVIDIA is best known as the architect of the GPU ecosystem that powers modern AI. Its H100 Tensor Core GPU, released in 2022, became the gold standard for training large language models, and that position has been extended dramatically by the Blackwell platform, which was rolled out through 2024 and 2025. As of April 2026, Blackwell systems are sold out through mid-year, with each GPU commanding approximately $40,000, and the architecture supports advanced features like 4-bit floating point inference, making it essential for training and running large AI models [3]. The next-generation Vera Rubin system, due in the second half of 2026, promises even greater leaps: Vera Rubin will use about twice as much power as Blackwell but will be far more efficient, delivering 10 times more performance per watt, and is made up of 1.3 million components [4]. Alongside its hardware ambitions, NVIDIA recently acquired Groq — a leading AI inference chip startup — and is pressing into the laptop processor market. NVIDIA unveiled its N1 and N1X series of AI-powered laptop processors, representing the company’s first comprehensive push into the consumer System-on-a-Chip (SoC) market, integrating high-performance Arm-based central processing units with its Blackwell graphics architecture [5].

NVIDIA matters because it does not merely sell chips — it sells an ecosystem. The CUDA software platform, developed over fifteen years, creates what analysts describe as a “compute moat”: the vast majority of AI frameworks, models, and workflows are written to run natively on CUDA, making switching costs extremely high. NVIDIA dominates 85% of the AI chip market, with analysts projecting $1 trillion in revenue by 2027 through AI infrastructure expansion and the Blackwell platform [6]. The dominance of Blackwell has created a massive compute moat for the industry’s largest players, with Microsoft and Meta both utilizing massive Blackwell clusters for their most advanced AI model deployments [7]. No other company, as of April 2026, simultaneously controls the hardware, interconnect fabric (NVLink), and software stack for AI training at scale. The only credible threats come from custom ASIC designers eating into specific workloads and AMD’s growing GPU portfolio — but neither has dislodged NVIDIA’s structural grip.

2. TSMC (Taiwan Semiconductor Manufacturing Company)

Taiwan Semiconductor Manufacturing Company — universally known as TSMC — is headquartered in Hsinchu, Taiwan, and was founded in 1987 by Morris Chang with backing from the Taiwanese government. It is not a chip designer in the conventional sense; rather, it is the world’s dominant contract foundry, the company that physically manufactures the chips that everyone else designs. Its customers include essentially every major name in this list: NVIDIA, AMD, Google, Broadcom, Apple, and AWS all depend on TSMC’s factories to turn silicon into semiconductors.

TSMC is best known for its mastery of advanced process nodes — the cutting-edge manufacturing techniques measured in nanometers that determine how powerful and efficient a chip can be. As of 2026, TSMC controls more than 90% of the advanced-node market required for cutting-edge AI chips, and for companies operating at scale, there is effectively no alternative [9]. In Q3 2025, the company captured a 72% share of the global semiconductor foundry market, outpacing the broader industry’s 17% growth with a 40% year-on-year revenue surge, with advanced technologies at nodes below 7 nanometers representing nearly 74% of TSMC’s wafer revenue [10,12]. In 2026, the company is moving aggressively into 2-nanometer production and has introduced its A16 node. TSMC plans to invest between $52 billion and $56 billion in 2026, a 27–37% increase from about $40.9 billion in 2025, with around 70–80% of that investment directed toward advanced process technologies such as 3-nanometer and 2-nanometer nodes [10]. Revenue consensus for 2026 is projected at approximately $160 billion, reflecting roughly 30% year-over-year growth.

TSMC matters because it is the unavoidable chokepoint of the entire AI chip supply chain. It is not merely a supplier but the singular gatekeeper to leading-edge compute. Even if another foundry undercuts TSMC’s pricing, the stakes are so high with AI that a chip company may pay a premium for the manufacturing certainty TSMC offers [12]. Furthermore, TSMC’s CoWoS advanced packaging technology — which enables the stacking and integration of multiple chiplets — is essential to the very architecture of modern AI accelerators like Blackwell and Google’s Ironwood. Intel’s most advanced node (Intel 18A) is targeting comparable capability to TSMC’s 3nm — a gap of 2–3 years from a process technology standpoint — and Samsung Foundry has 3nm in production but at lower yields and with fewer premium customers [11]. Geopolitical risk around Taiwan remains the single most discussed systemic threat to the global AI hardware supply chain, which is why the United States and Japan are investing heavily to encourage TSMC to build fabs abroad — efforts already well underway in Arizona.

3. Broadcom

Broadcom Inc. is headquartered in San Jose, California, and is led by CEO Hock Tan, who has transformed what was once a diversified networking and storage chip company into the indispensable architect of custom AI silicon for the world’s largest technology companies. Broadcom operates through two primary segments: Semiconductor Solutions and Infrastructure Software (the latter bolstered by its $69 billion acquisition of VMware in 2023).

Broadcom is best known in the AI chip context for designing application-specific integrated circuits, or ASICs — custom chips purpose-built to a single customer’s exact AI workload. Broadcom designs custom AI accelerator chips (called XPUs) for the world’s largest technology companies, including Google, Meta, OpenAI, and Anthropic. These chips are optimized for AI training and inference workloads and, unlike Nvidia’s general-purpose GPUs, are tailored to each customer’s specific AI model architectures, delivering superior performance-per-watt for targeted workloads. Broadcom reported $8.4 billion in AI semiconductor revenue for Q1 FY2026, ended February 2026, representing a 106% year-over-year increase [13]. Its most prominent product is Google’s Tensor Processing Unit (TPU), co-developed with Google for over a decade. Google delivered its first custom ASICs back in 2015, with its TPUs originally designed for standard cloud computing workloads, and both Google and Amazon relied on Broadcom to help develop their silicon [16]. More recently, the company cemented a landmark deal with Meta — committing 1 gigawatt of custom chips — and in late 2025 signed a massive multi-year partnership with OpenAI to co-develop and deploy 10 gigawatts of custom AI accelerators [15, 18]. Broadcom is the lead design partner for Google’s TPU v7 “Ironwood” and is expanding its work with Meta on their MTIA accelerators [17].

Broadcom matters because it represents the most credible structural alternative to NVIDIA’s dominance. As hyperscalers grow large enough to justify the enormous upfront cost of custom silicon, they are increasingly turning to Broadcom to reduce their dependence on NVIDIA and cut total cost of ownership. Counterpoint Research projects that AI server compute ASIC shipments among the top 10 hyperscalers will triple between 2024 and 2027, fueled by surging demand for Google’s TPU infrastructure, AWS Trainium clusters, and Meta’s MTIA accelerators [14]. According to TrendForce, custom ASIC shipments from cloud providers are projected to grow 44.6% in 2026, while GPU shipments are expected to grow 16.1%, signaling a decisive shift in the AI hardware landscape as hyperscalers increasingly invest in their own silicon [20]. Broadcom’s networking chips — particularly the Tomahawk 6 switching chip, capable of 102.4 Tbps — are also the standard for connecting AI GPU clusters, giving it dual relevance in both compute and the fabric that connects compute [17].

4. Google (Alphabet)

Google’s parent company Alphabet is headquartered in Mountain View, California, and its AI chip efforts are the most mature in-house silicon program among the hyperscalers. Google began designing its own Tensor Processing Units in 2013, driven by the realization that the explosive growth of neural network inference inside its products would require hardware purpose-built for matrix math, not the general-purpose processors sold commercially.

Google is best known for its TPU line, now in its seventh generation. Ironwood, Google’s seventh-generation TPU, is specifically designed for inference, built to handle the massive computational demands of thinking models like large language models and mixture-of-experts architectures, and scales up to 9,216 chips, offering 42.5 exaflops of compute power and making it more powerful than the world’s largest supercomputer [21]. The Ironwood chip is designed from the ground up for the “age of inference” — the current era in which AI deployment, not training, is the dominant computational workload. Ironwood offers a 10X peak performance improvement over TPU v5p and more than 4X better performance per chip for both training and inference workloads compared to the prior generation Trillium, making it Google’s most powerful and energy-efficient custom silicon to date [22]. With 4,614 TFLOPs of FP8 compute capability, Ironwood surpasses even Nvidia’s latest Blackwell GB200 GPU on raw inference performance, and its smaller physical footprint enables higher rack density, lowering the total cost of ownership for hyperscale deployments [25].

Google matters because it represents what every other large tech company is trying to become: an AI company that is not beholden to NVIDIA for its core infrastructure. TPUs power Google Search, Gmail, YouTube, Google Maps, and Google’s Gemini models — products used by billions of people daily. Beyond its own walls, Google is now selling TPU access commercially through Google Cloud, making Ironwood available to any enterprise that wants to reduce its GPU costs. Anthropic, in its collaboration with Google, aims to access up to one million TPU chips — a deal valued at tens of billions of dollars — with this arrangement set to give Anthropic over a gigawatt of computing capacity by 2026 [2]. The company also unveiled its quantum chip Willow in December 2024 and is pursuing Project Suncatcher, a constellation of solar-powered satellites equipped with TPUs, signaling ambitions that extend well beyond conventional chip competition [2].

5. AMD (Advanced Micro Devices)

Advanced Micro Devices is headquartered in Santa Clara, California, and has undergone one of the most remarkable corporate transformations in recent semiconductor history, evolving under CEO Lisa Su from a struggling also-ran behind Intel into the world’s second-largest AI chip company. AMD is a fabless designer like NVIDIA, also relying primarily on TSMC for manufacturing.

AMD is best known in the AI space for its Instinct series of GPU accelerators, which have matured from credible alternatives into genuine competitive threats. The AMD Instinct MI350 Series GPUs represent the fastest-ramping product in company history, already deployed at scale by leading cloud providers including Oracle Cloud Infrastructure, and the upcoming “Helios” systems with AMD Instinct MI450 Series GPUs are expected to deliver rack-scale performance leadership with industry-leading memory capacity and scale-out bandwidth beginning in the third quarter of 2026, followed by the MI500 Series [26]. The MI350 secured major commitments before it even fully ramped, with Microsoft, Meta, and OpenAI all making deployment commitments that validate AMD’s competitive positioning [29]. On the software side, AMD has invested heavily in ROCm, its open-source alternative to NVIDIA’s CUDA, with the explicit goal of making it a viable development platform across the ecosystem. If you scan the world’s fastest supercomputers, AMD silicon is everywhere — Frontier and El Capitan hold the number one and number two slots, proof of its ability to scale compute at absurd levels [30].

AMD matters for two interconnected reasons. First, it is the only company in the world that can supply high-volume, general-purpose AI accelerators at data center scale as an alternative to NVIDIA — a fact that gives hyperscalers and cloud providers critical negotiating leverage and supply diversification. While overtaking NVIDIA remains unlikely in the near term, AMD’s growth trajectory could deliver outsized returns for investors if it captures even a modest additional slice of the market, and for enterprises and cloud providers, the competition benefits customers through innovation, pricing pressure, and diversified supply [8]. Second, AMD is simultaneously strong in server CPUs with its EPYC line, which increasingly handles inference workloads at scale — meaning AMD competes across the full AI compute stack, not just in accelerators. Analysts estimate AMD’s AI GPU revenue could reach between $10 and $12 billion in 2026, driven by data center growth [28], representing enormous growth from just a few years prior.

6. Amazon Web Services (AWS)

Amazon Web Services is a subsidiary of Amazon.com and is headquartered in Seattle, Washington. It is the world’s largest cloud computing platform by revenue, and its foray into custom AI silicon stems directly from a strategic imperative: reduce dependence on NVIDIA’s expensive and often-backordered GPUs, lower costs for customers, and protect competitive positioning against Microsoft Azure and Google Cloud, both of which have their own proprietary chips.

AWS is best known in the AI hardware space for its Trainium family of training accelerators and its Inferentia chips for inference. Trainium3 UltraServers deliver high performance for AI workloads with up to 4.4x more compute performance, 4x greater energy efficiency, and almost 4x more memory bandwidth than Trainium2 UltraServers, scaling up to 144 Trainium3 chips and delivering up to 362 FP8 PFLOPs with 4x lower latency [33]. The economic case for Trainium is compelling: early testing shows companies can save up to 50% on costs compared to GPU training, providing enterprises ramping up AI adoption with a balance sheet-saving option [32]. The customer endorsements are also striking: two of the world’s leading AI labs — Anthropic and OpenAI — are committed to Trainium, with Anthropic naming AWS its primary training partner and OpenAI committing to consume 2 gigawatts of Trainium capacity through AWS infrastructure [34].

AWS matters because it operates at a scale that few can match, and its chips are deeply integrated into Amazon Bedrock, the managed AI model service that underpins a large and rapidly growing share of enterprise AI deployments globally. Trainium2 handles the majority of the inference traffic on Amazon’s Bedrock service, which supports the building of AI applications by Amazon’s many enterprise customers, with one AWS executive calling Bedrock potentially as large as EC2 one day [35]. Unlike Broadcom’s chips, which are custom-designed for specific customers, and unlike NVIDIA and AMD’s chips, which are sold openly, AWS’s Trainium occupies a unique middle position: built internally and deployed at hyperscale, but also made available to any AWS cloud customer. It is, in essence, a cost-reduction strategy that has become a product.

7. Qualcomm

Qualcomm is headquartered in San Diego, California, and was founded in 1985 by Irwin Jacobs and six co-founders. For most of its history, Qualcomm was synonymous with wireless communication technology and mobile chip design — it invented key elements of 3G and 4G standards and collected licensing royalties on a vast portion of the world’s cellular devices. But in the AI era, Qualcomm is staking a claim to a fundamentally different kind of dominance: the AI that lives on your device, not in the cloud.

Qualcomm is best known for its Snapdragon platform, a system-on-chip that powers the majority of Android flagship smartphones and, increasingly, Windows laptops. Each Snapdragon chip includes a dedicated Hexagon neural processing unit (NPU) designed for on-device AI inference. At CES 2026, Qualcomm announced the Snapdragon X2 Plus for PCs, featuring an 80 TOPS Hexagon NPU critical for enabling agentic experiences — where the PC can perform complex, proactive actions on behalf of the user [37]. The company has expanded its ambitions well beyond smartphones: Qualcomm’s automotive revenue hit $1.1 billion in its fiscal Q1 2026, up 15% year over year, and the company signed a long-term collaboration with Neura Robotics to jointly develop “brain and nervous system” reference architectures for next-generation humanoid robots [36]. CEO Cristiano Amon has made the strategic thesis explicit, arguing at both Web Summit 2026 and Davos 2026 that the AI race will be won at the edge, where devices, data, and users converge, with success depending on constant reinvention, efficient on-device computing, and competition across both edge and data center inference [36].

Qualcomm matters because the AI compute war has two very different fronts, and Qualcomm owns one of them. While NVIDIA, Broadcom, Google, AMD, and AWS compete for data center supremacy, Qualcomm is positioning itself as the indispensable silicon layer for the billions of edge devices — smartphones, PCs, automobiles, industrial machines, wearables, and robots — where AI will eventually run locally. On-device inference means lower latency, stronger privacy, and no cloud costs, all of which are becoming decisive competitive factors. Within less than a week of its release, DeepSeek R1-distilled models were running on PCs and smartphones powered by Snapdragon platforms, demonstrating the maturity of edge AI inference and accelerating demand for powerful chips at the edge [38]. As AI models continue to shrink through distillation and quantization, the relative importance of Qualcomm’s edge computing platform will grow proportionally.

8. Intel

Intel Corporation is headquartered in Santa Clara, California, and was founded in 1968 by Gordon Moore and Robert Noyce. For more than four decades, Intel was the most powerful company in the semiconductor industry — its x86 processors powered virtually every PC and server on the planet. The rise of AI, however, exposed a fundamental strategic misalignment: Intel had built its empire on sequential, general-purpose computing, whereas AI demands massively parallel, matrix-heavy workloads that GPUs handle far better.

Intel is best known in the AI chip context for its Gaudi series of accelerators, inherited through its 2019 acquisition of Habana Labs. The Gaudi 3 chip trains models 1.5 times faster, outputs results 1.5 times faster, and uses less power than Nvidia’s H100 chip [1]. However, Gaudi has badly underperformed commercial expectations: the Gaudi disappointment underscores Intel’s persistent AI travails, years after it declined to pick a single strategy that could counter its skyrocketing rival Nvidia [41]. Intel subsequently cancelled its planned Falcon Shores GPU successor, designating it an internal test chip, and is now betting its future on Jaguar Shores — a rack-scale AI solution targeting volume production in the second half of 2026. Jaguar Shores is to be built on Intel’s 18A process node, feature HBM4 memory from SK Hynix, and use silicon photonics for high-speed interconnects, disaggregating compute and memory resources across an entire rack for better efficiency [40].

Intel matters — even in its current weakened state — for several structural reasons. It remains the dominant supplier of x86 server CPUs, and those CPUs sit inside virtually every AI data center today, handling coordination, preprocessing, and the growing volume of CPU-based inference workloads. Its Intel Foundry Services business, centered on the 18A process node, could, if successful, provide the Western world with a domestic advanced chip manufacturing alternative to TSMC — a geopolitical imperative of the first order. Intel co-CEO Michelle Johnston Holthaus stated: “This is an attractive market for us over time, but I am not happy with where we are today. On the one hand, we have a leading position as the host CPU for AI servers, and we continue to see a significant opportunity for CPU-based inference on-prem and at the edge. On the other hand, we’re not yet participating in the cloud-based AI datacenter market in a meaningful way” [39]. Intel is ranked eighth here not because it is unimportant, but because its AI accelerator business has consistently failed to match its ambitions — and until Jaguar Shores ships at volume with credible software support, that gap will remain.

References

1. “10 top AI hardware and chip-making companies in 2026,” TechTarget — https://www.techtarget.com/searchdatacenter/tip/Top-AI-hardware-companies
2. “15 Leading AI Hardware Companies Dominating the Market in 2026,” Big Data Supply, Inc. — https://bigdatasupply.com/leading-ai-hardware-companies/
3. “Nvidia Stock Analysis 2026: $1 Trillion AI Demand,” Intellectia AI — https://intellectia.ai/blog/nvidia-stock-analysis-2026-ai-demand
4. “First look at Nvidia’s AI system Vera Rubin and how it beats Blackwell,” CNBC — https://www.cnbc.com/2026/02/25/first-look-at-nvidias-ai-system-vera-rubin-and-how-it-beats-blackwell.html
5. “The Silicon Power Play: Nvidia’s New AI Laptop Chips,” FinancialContent — https://www.financialcontent.com/article/marketminute-2026-2-23-the-silicon-power-play-nvidias-new-ai-laptop-chips-signal-a-high-stakes-grab-for-the-pc-market
6. “Nvidia’s Blackwell Platform Set to Lock in AI Inference Dominance Before May 2026,” AInvest — https://www.ainvest.com/news/nvidia-blackwell-platform-set-lock-ai-inference-dominance-2026-2604/
7. “The Blackwell Era: NVIDIA’s 30x Performance Leap Ignites the 2026 AI Revolution,” FinancialContent — https://markets.financialcontent.com/wral/article/tokenring-2026-1-12-the-blackwell-era-nvidias-30x-performance-leap-ignites-the-2026-ai-revolution
8. “NVIDIA vs AMD 2026: AI Chip Showdown,” IBTimes Australia — https://www.ibtimes.com.au/nvidia-vs-amd-2026-ai-chip-showdown-who-dominates-data-centers-blackwell-mi400-battle-1866177
9. “TSMC Stock: In the AI Arms Race, The Foundry Wins,” Trefis — https://www.trefis.com/stock/tsmc/articles/588231/tsmc-stock-in-the-ai-arms-race-the-foundry-wins/2026-01-22
10. “Will TSM’s Aggressive Capex Plan Strengthen Its Foundry Dominance?” Yahoo Finance/Zacks — https://finance.yahoo.com/sectors/technology/articles/tsms-aggressive-capex-plan-strengthen-134200418.html
11. “TSMC Q1 2026: $35.7B Record Revenue — The AI Chip Chokepoint That Controls Everything,” Vucense — https://vucense.com/ai-intelligence/industry-business/tsmc-q1-2026-record-revenue-ai-chip-chokepoint/
12. “TSMC’s Structural Dominance: Why the 2026 AI Infrastructure Winner is the Foundry, Not the Chipmaker,” AInvest — https://www.ainvest.com/news/tsmc-structural-dominance-2026-ai-infrastructure-winner-foundry-chipmaker-2512/
13. “Broadcom AI Revenue Surges 106%: Custom Chip Strategy 2026,” Tech-Insider.org — https://tech-insider.org/broadcom-ai-revenue-custom-chips-2026/
14. “Broadcom Set To Dominate Custom AI Chip Market With 60% Share By 2027,” Yahoo Finance/Counterpoint — https://finance.yahoo.com/news/broadcom-set-dominate-custom-ai-163116560.html
15. “Meta Broadcom AI Chip Deal 2026: 1GW MTIA, 2nm,” Oplexa — https://oplexa.com/meta-broadcom-ai-chip-deal-2026/
16. “Meta doubles down on partnership with Broadcom, committing to 1 gigawatt of custom AI processors,” SiliconANGLE — https://siliconangle.com/2026/04/14/meta-doubles-partnership-broadcom-committing-1-gigawatt-custom-ai-processors/
17. “Broadcom (AVGO) in 2026: The Industrial Architect of the AI Era,” FinancialContent — https://www.financialcontent.com/article/finterra-2026-4-15-broadcom-avgo-in-2026-the-industrial-architect-of-the-ai-era
18. “OpenAI and Broadcom to co-develop 10GW of custom AI chips,” Tom’s Hardware — https://www.tomshardware.com/openai-broadcom-to-co-develop-10gw-of-custom-ai-chips
19. “Broadcom, Anthropic & Google: Developing Custom AI Chips,” Manufacturing Digital — https://manufacturingdigital.com/articles/broadcom-anthropic-google-developing-custom-ai-chips
20. “Top 20+ AI Chip Makers: NVIDIA & Its Competitors,” AImultiple — https://aimultiple.com/ai-chip-makers
21. “Ironwood: The first Google TPU for the age of inference,” Google Blog — https://blog.google/innovation-and-ai/infrastructure-and-cloud/google-cloud/ironwood-tpu-age-of-inference/
22. “Ironwood TPUs and new Axion-based VMs for your AI workloads,” Google Cloud Blog — https://cloud.google.com/blog/products/compute/ironwood-tpus-and-new-axion-based-vms-for-your-ai-workloads
23. “Google’s rolling out its most powerful AI chip, taking aim at Nvidia,” CNBC — https://www.cnbc.com/2025/11/06/google-unveils-ironwood-seventh-generation-tpu-competing-with-nvidia.html
24. “Google TPU Architecture: 7 Generations Explained,” Introl Blog — https://introl.com/blog/google-tpu-architecture-complete-guide-7-generations
25. “Google TPU Chip Ironwood Technology Explained,” Nevsemi Electronics — https://www.nevsemi.com/blog/google-tpu-chip-ironwood-technology-explained
26. “AMD Unveils Strategy to Lead the $1 Trillion Compute Market,” AMD Newsroom — https://www.amd.com/en/newsroom/press-releases/2025-11-11-amd-unveils-strategy-to-lead-the-1-trillion-compu.html
27. “AMD MI400 Series: $7.2B AI GPU Challenging Nvidia [2026],” Tech-Insider.org — https://tech-insider.org/amd-mi400-series-ai-gpu-data-center-2026/
28. “AMD’s new Instinct AI GPUs will possibly bring in up to $12 billion in revenue for 2026,” Tweaktown — https://www.tweaktown.com/news/105790/amds-new-instinct-ai-gpus-will-possibly-bring-in-up-to-12-billion-revenue-for-2026/index.html
29. “AMD’s MI350: The AI Accelerator That Could Challenge Nvidia’s Dominance In 2026,” Seeking Alpha — https://seekingalpha.com/article/4856532-amds-mi350-ai-accelerator-that-could-challenge-nvidias-dominance-in-2026
30. “13 Top AI Chip Companies You Should Know About,” VKTR — https://www.vktr.com/ai-market/10-top-ai-chip-companies/
31. “Amazon releases an impressive new AI chip,” TechCrunch — https://techcrunch.com/2025/12/02/amazon-releases-an-impressive-new-ai-chip-and-teases-a-nvidia-friendly-roadmap/
32. “AWS Launches Trainium3 Chip to Challenge Nvidia,” Data Center Knowledge — https://www.datacenterknowledge.com/data-center-chips/aws-launches-tranium3-chip-to-challenge-nvidia-ai-dominance
33. “Trainium3 UltraServers now available,” About Amazon — https://www.aboutamazon.com/news/aws/trainium-3-ultraserver-faster-ai-training-lower-cost
34. “AWS and Cerebras Collaboration Aims to Set a New Standard for AI Inference Speed,” AWS Press Center — https://press.aboutamazon.com/aws/2026/3/aws-and-cerebras-collaboration-aims-to-set-a-new-standard-for-ai-inference-speed-and-performance-in-the-cloud
35. “An exclusive tour of Amazon’s Trainium lab,” TechCrunch — https://techcrunch.com/2026/03/22/an-exclusive-tour-of-amazons-trainium-lab-the-chip-thats-won-over-anthropic-openai-even-apple/
36. “Qualcomm’s CEO Says the Winner of Edge AI Will Win the Entire AI Race,” The Motley Fool — https://www.fool.com/investing/2026/04/11/qualcomms-ceo-says-the-winner-of-edge-ai-will-win/
37. “Qualcomm Unveils Future of Intelligence at CES 2026,” Futurum Group — https://futurumgroup.com/insights/qualcomm-unveils-future-of-intelligence-at-ces-2026-pushes-the-boundaries-of-on-device-ai/
38. “AI Disruption is Driving Innovation in On-device Inference,” Edge AI and Vision Alliance — https://www.edge-ai-vision.com/2025/02/ai-disruption-is-driving-innovation-in-on-device-inference/
39. “Intel Shifts AI Strategy as Falcon Shores Falters, Eyes Jaguar Shores for 2026,” VoIP Review — https://voip.review/2025/02/04/intel-shifts-ai-strategy-falcon-shores-falters-eyes-jaguar-shores-2026/
40. “Intel Jaguar Shores AI: Betting Big on Rack-Scale and 18A Process,” OfZenAndComputing — https://www.ofzenandcomputing.com/intel-jaguar-shores-ai/
41. “Intel AI Chip Struggles — Revenue Projections Fall,” GBAF — https://www.globalbankingandfinance.com/a-year-on-intels-touted-ai-chip-deals-have-fallen-short/

###

Filed under: Uncategorized |