NVIDIA and Arm Holdings stand out as two contrasting AI plays. NVIDIA reported FY2026 revenue of $215.9 billion with a 55.6% net margin, while Arm earned $4.9 billion in revenue with an 18.4% net margin. NVIDIA's acquisition of Hugging Face for $12.9 billion signals a push into AI developer tools, and the company counts just two customers responsible for 22% and 14% of its revenue. Arm, meanwhile, licenses the architecture behind 99% of global smartphones but faces headwinds from open-source alternatives and dependence on the mobile market.
CoreWeave, a major NVIDIA GPU customer, earned $5.1 billion in revenue in FY2025 — a 168% increase — but Microsoft accounted for roughly 67% of its business. The company posted a $1.2 billion net loss, highlighting how quickly growth can come with heavy costs. D-Wave Quantum, which serves clients like Mastercard, Pfizer, and BASF, saw revenue rise 179% to $24.6 million but recorded a $355.1 million net loss.
On the chip front, Marvell Technology unveiled the industry's first 2nm optical interconnects for AI data centers, capable of 102.4Tbps co-packaged optics. Marvell posted record Q2 FY2027 revenue of $2.739 billion, with Data Center revenue up 46% year over year to $2.17 billion. CEO Matt Murphy noted that scale-up optics demand is growing more than 10 times faster than current networks, and the company expanded partnerships with both Google and NVIDIA.
Chinese AI lab StepFun launched Step 5 Preview, a sparse Mixture-of-Experts model with 600 billion total parameters but only 27 billion activated per token. It scored 44 on the Artificial Analysis Intelligence Index, beating Gemini 3.8 Flash's score of 41, and costs $0.72 per task — about 42% less than Gemini's $1.24. API access opened September 20, with open weights scheduled for October 15, 2026.
Big Tech companies are preparing for new accounting transparency around AI spending. Under new FASB guidance called DISE, Alphabet and Meta will need to break down costs for staff salaries and asset depreciation. Alphabet's R&D spending hit $18.2 billion in Q2, while Meta's reached $21.6 billion. Investors have pushed for clearer disclosure on ballooning AI capital expenditures, and these details will appear in financial filing footnotes.
Amazon and Alphabet are set to spend roughly $400 billion on AI infrastructure in 2026. AWS is growing 37% year over year, while Google Cloud surged 82%. Amazon's Andy Jassy noted that data-center servers take nearly three years to break even, meaning much of the 2026 build won't fully monetize until 2028. Alphabet's vertical integration with TPUs and Gemini models gives it an edge in operating leverage, and the company trades at a forward P/E of 23 compared to Amazon's negative $7.6 billion in trailing free cash flow.
Enterprise AI strategy is shifting toward data retention as a key differentiator. Companies including NVIDIA, Palantir, and Novo Nordisk are drawing clear lines on which AI systems they will use. Privacy architecture is emerging as a competitive advantage, with providers that keep sensitive data under customer control poised to win more enterprise contracts. Finance and security leaders now evaluate which models can perform best within existing controls.
Asian investors are demanding proof that AI spending will translate into revenue and earnings. A Bank of America survey found that the region's investors increasingly focus on monetization rather than pure technology exposure. Chris Oberoi, head of Asia-Pacific research at Bank of America, said fundamentals drive everything, reflecting a more mature investment phase for AI across the region.
NFL legend Tom Brady backed a $200 million funding round for eMed, a digital healthcare startup using AI for employee wellness. The Miami-based company's valuation now exceeds $2 billion after the Series A financing. eMed offers access to GLP-1 medications, at-home blood testing, and ongoing clinical support for employers, though its growth will depend on delivering effective clinical outcomes.
Lawmakers Valerie Foushee and James Moylan introduced the SAFE Platforms Act, a bipartisan bill targeting online scams that exploit AI. The legislation would require platforms to deploy AI-powered detection tools and create a national database to track scam information. Consumer advocacy groups praised the bill, though some critics raised free speech and content moderation concerns.
Key Takeaways
- NVIDIA reported FY2026 revenue of $215.9 billion with a 55.6% net margin and acquired Hugging Face for $12.9 billion, while Arm earned $4.9 billion with an 18.4% net margin and competes with Intel and AMD.
- CoreWeave earned $5.1 billion in FY2025 revenue (up 168%) but Microsoft accounted for about 67% of its revenue and the company posted a $1.2 billion net loss.
- Marvell unveiled the industry's first 2nm optical interconnects for AI data centers at 102.4Tbps, with Data Center revenue up 46% year over year to $2.17 billion.
- StepFun's Step 5 Preview model scores 44 on the Artificial Analysis Intelligence Index, beating Gemini 3.8 Flash's 41, and costs 42% less at $0.72 per task.
- Alphabet's R&D spending reached $18.2 billion in Q2 and Meta's hit $21.6 billion, with new FASB DISE guidance requiring detailed breakdowns of AI costs.
- Amazon and Alphabet plan roughly $400 billion in AI infrastructure spending in 2026, with Google Cloud surging 82% year over year and AWS growing 37%.
- Companies including NVIDIA, Palantir, and Novo Nordisk are prioritizing data retention policies and privacy architecture as competitive advantages in enterprise AI.
- Asian investors increasingly focus on AI monetization over technology exposure, according to a Bank of America survey, with fundamentals driving investment decisions.
- Tom Brady backed a $200 million Series A round for eMed, an AI healthcare startup whose valuation now exceeds $2 billion.
- The bipartisan SAFE Platforms Act would require online platforms to use AI-powered tools to detect and prevent scams, creating a national scam-tracking database.
Arm Holdings vs NVIDIA as top AI stock picks for 2026
Arm Holdings and NVIDIA are compared as leading AI semiconductor stocks. Arm earned $4.9 billion in revenue in FY2026 with an 18.4% net margin, while NVIDIA reported $215.9 billion in revenue with a 55.6% net margin. NVIDIA acquired Hugging Face for $12.9 billion and faces U.S. export controls affecting China sales. Arm competes with Intel and AMD while licensing architecture used in 99% of global smartphones.
CoreWeave vs D-Wave Quantum as AI and quantum computing stocks
CoreWeave and D-Wave Quantum are compared as high-growth computing stocks. CoreWeave earned $5.1 billion in revenue in FY2025, a 168% increase, but reported a $1.2 billion net loss and Microsoft accounted for about 67% of revenue. D-Wave Quantum earned $24.6 million in FY2025, up 179%, while recording a $355.1 million net loss. CoreWeave depends heavily on NVIDIA GPUs while D-Wave serves clients like Mastercard, Pfizer, and BASF.
Arm Holdings vs NVIDIA as better AI stock choice
Arm Holdings and NVIDIA are compared as leading AI chip stocks for investors. Arm reported FY2026 revenue of $4.9 billion with $904 million in net income and a 6.0x current ratio. NVIDIA reported FY2026 revenue of $215.9 billion, up 65.5%, with $120.1 billion in net income and a 55.6% net margin. NVIDIA acquired Hugging Face for $12.9 billion and has two customers representing 22% and 14% of revenue. Arm faces competition from Intel and AMD.
Arm Holdings vs NVIDIA as better AI stock choice
Arm Holdings and NVIDIA are compared as leading AI semiconductor stocks for 2026. Arm earned $4.9 billion in FY2026 revenue with an 18.4% net margin and $979 million in free cash flow. NVIDIA earned $215.9 billion in FY2026 revenue, up 65.5%, with a 55.6% net margin and $96.7 billion in free cash flow. NVIDIA acquired Hugging Face for $12.9 billion and has agreements with Microsoft. Arm faces risks from open-source architectures and mobile market dependence.
Arm Holdings vs ASML as better AI stock pick
Arm Holdings and ASML are compared as leading AI-related stocks. Arm licenses energy-efficient architecture used in over 350 billion chips and earned $4.9 billion in FY2026 revenue with an 18.4% net margin. ASML is the sole provider of extreme ultraviolet lithography machines and earned about 32.7 billion euros in FY2025 revenue with a 29.4% net margin. Arm faces competition from RISC-V while ASML faces export restrictions to China and cyclical industry risks.
StepFun launches Step 5 Preview AI model beating Gemini on cost
Chinese AI lab StepFun released Step 5 Preview, a flagship model for agentic work. It scores 44 on the Artificial Analysis Intelligence Index, beating Gemini 3.8 Flash's 41. The model costs $0.72 per task, about 42% less than Gemini's $1.24. It uses a sparse Mixture-of-Experts architecture with 600 billion total parameters and 27 billion activated per token. API access opened on September 20, with open weights scheduled for October 15, 2026.
StepFun launches 600B-parameter MoE model Step 5 Preview
StepFun released Step 5 Preview, a sparse Mixture-of-Experts model with 600 billion total parameters and 27 billion activated per token. It offers a 1 million-token context window and accepts text, image, and video input. API pricing is $1.00 per million input tokens and $2.70 per million output tokens. The model scored 44 on Artificial Analysis Intelligence Index and ran at high effort. Open weights are scheduled for October 15, 2026.
Enterprise AI value depends on data retention policies
The most important benchmark for enterprise AI is no longer just model intelligence but data retention policies. Companies like Nvidia, Palantir, and Novo Nordisk are drawing lines on which AI systems they will use. Privacy architecture could become AI's next competitive advantage. Providers that deliver frontier capabilities while keeping sensitive data under customer control may win more enterprise contracts. Finance and security leaders now ask which models can perform best within existing controls.
Asian AI investors demand revenue proof ahead of Xi-Trump summit
Asian investors are demanding clearer evidence that artificial intelligence investments will generate revenue and earnings. A Bank of America survey found that investors increasingly focus on monetization rather than just technology exposure. Chris Oberoi, head of Asia-Pacific research at Bank of America, said fundamentals drive everything. This shift reflects AI moving into a more mature investment phase across the region.
Big Tech prepares new accounting rules for AI spending
Companies like Alphabet and Meta are preparing to disclose detailed breakdowns of their AI spending under new Financial Accounting Standards Board guidance. Alphabet's research and development spending totaled $18.2 billion in Q2, while Meta's reached $21.6 billion in the same period. The new standard called DISE requires companies to break down costs for staff salaries and asset depreciation. Investors have pushed for more transparency on ballooning AI capital expenditures. The disclosures will appear in financial filing footnotes.
Marvell unveils industry-first 2nm optical chips for AI data centers
Marvell Technology demonstrated the world's first 2nm optical interconnects for AI data centers, including 102.4Tbps co-packaged optics. The company reported record Q2 FY2027 revenue of $2.739 billion, with Data Center revenue reaching $2.17 billion, up 46% year over year. CEO Matt Murphy said scale-up optics demand is growing more than 10 times faster than current networks. Marvell also expanded partnerships with Google and NVIDIA, and its Investor Day is scheduled for October 6, 2026.
Hong Kong deputy police chief warns AI is a new national security threat
Retiring Hong Kong deputy police commissioner Andrew Kan Kai-yan said artificial intelligence poses a new national security threat. He spoke about using AI to counter AI-driven risks and the ongoing challenge of soft resistance. Chief Executive John Lee Ka-chiu announced plans to adopt an AI risk-governance strategy to combat criminal use of the technology. Kan served 35 years on the force and headed the National Security Department.
What to expect from Amazon and Alphabet's AI spending in 2027
Amazon and Alphabet are set to spend roughly $400 billion on AI infrastructure in 2026, with AWS growing 37% and Google Cloud surging 82% year over year. Amazon's Andy Jassy said data-center servers take nearly three years to break even, so much of the 2026 build will not fully monetize until 2028. Alphabet trades at a forward P/E of 23 while Amazon has negative $7.6 billion in trailing free cash flow. Alphabet's vertical integration with TPUs and Gemini models gives it an edge in operating leverage.
Bipartisan bill targets AI-powered online scams
Lawmakers Valerie Foushee and James Moylan introduced the SAFE Platforms Act, a bipartisan bill targeting online scams that exploit AI. The bill would require online platforms to use AI-powered tools to detect and prevent scams. It would also create a national database to track and share scam information. Consumer advocacy groups praised the bill, though some critics raised concerns about free speech and content moderation.
Tom Brady's investment helps eMed AI healthcare startup reach $2 billion valuation
NFL legend Tom Brady backed a $200 million funding round for eMed, a digital healthcare company using AI for employee wellness. The Miami-based startup's valuation now exceeds $2 billion after the Series A financing. eMed offers access to GLP-1 medications, at-home blood testing, and ongoing clinical support for employers. The company's growth will depend on delivering effective clinical support to patients and employers.
Sources
- Better Artificial Intelligence Stock: Arm Holdings vs. NVIDIA
- CoreWeave vs. D-Wave Quantum: Is the AI Giant or the Quantum Computer Upstart the Better Stock Investment?
- Better Artificial Intelligence Stock: Arm Holdings vs. NVIDIA
- Better Artificial Intelligence Stock: Arm Holdings vs. NVIDIA
- Better Artificial Intelligence Stock: Arm Holdings vs. ASML
- China’s StepFun Releases Step 5 Preview, Beats Gemini 3.8 Flash On Performance And Cost
- StepFun Launches Step 5 Preview: A 600B-Total, 27B-Active MoE Model With 1M Context for Long-Horizon Agentic Work
- The Best Enterprise AI System Is One CFOs Are Allowed to Use
- AI investors in Asia seen shifting focus to earnings ahead of Xi-Trump summit
- Big Tech Braces for Accounting Rules Set to Demystify AI Costs
- Marvell Smashes the AI 'Power Wall' with Industry-First 2nm Optical Chips
- AI poses new national security threat, outgoing deputy police chief warns
- Here’s What to Expect From AI’s Two Biggest 2026 Spenders in 2027
- AI-powered online scams targeted in bipartisan bill - Live Updates
- Tom Brady's $200 million investment helped AI healthcare startup eMed cross $2 billion valuation
Comments
Please log in to post a comment.