Nvidia Microsoft Meta and AMD Drive AI Growth and Competition

Dan Loeb’s Third Point added Nvidia, Microsoft and Meta shares in the third quarter of 2026, alongside SpaceX. The companies point to different sources of AI growth: Nvidia says revenue per gigawatt rises from about $18 billion with Hopper to roughly $40 billion with Vera Rubin; Microsoft reports 48% Azure growth and an eightfold year-over-year increase in Copilot seats; and Meta’s revenue grew 26%, with ad prices up 14% and impressions up 10%. SpaceX is expanding its AI infrastructure business while pursuing rapid reuse of Starship.

Chip competition is intensifying as Nvidia faces a sell-off and questions about its leadership. AMD shares fell 21% over a month, versus a 40% decline for Nvidia. AMD denied allegations in a short-seller report that Reuters could not independently verify. Its third-quarter revenue rose 45% to $13.4 billion, including 97% growth in data-center revenue to $8.5 billion. Nvidia reported fiscal third-quarter revenue of $115.8 billion, up 103%, and forecast about $130 billion for the next quarter. Nvidia retains a connected hardware, software and networking ecosystem and supply advantages, but analysts say easier model portability could pressure its market share and margins. Anthropic says Claude Mythos can shift to alternative chips in days, down from six to 12 months; Alphabet’s TPUs and AMD’s GPUs are among the alternatives, while Taiwan Semiconductor Manufacturing could benefit from broader chip demand.

Broadcom is positioning itself in custom AI chips, including TPUs developed with Alphabet that can cost less than GPUs for some workloads. Anthropic has built much of its computing infrastructure around TPUs. Broadcom projects AI semiconductor revenue of $115 billion in 2027 and $230 billion in 2028; those are company forecasts, not guaranteed results. A separate investment analysis presents On Semiconductor as a less concentrated way to gain exposure to AI: its power and sensing chips serve data centers, industrial equipment and electric vehicles, and could support robots, drones and industrial machines. Analysts expect $2 billion in free cash flow in 2027 and $2.3 billion in 2028, though the company faces EV-cycle, competition and acquisition risks.

OpenAI says strengthened safeguards reduced adaptive attempts to extract protected reasoning from GPT-6.2 Astra and GPT-6.2 Sol from 54% to 0.2%; extraction tied to the current request fell from 22.9% to 0.9%. Its controls check requests, responses and tool-call arguments, while account-level investigations target coordinated campaigns. OpenAI says weekly enforcement actions in 2026 involved thousands to tens of thousands of accounts. Separately, ElevenLabs’ ElevenAI desktop app transcribes speech in more than 90 languages, records meetings without a bot, and produces summaries, speaker labels, action items and draft follow-ups. The article says audio is not stored and transcript history remains on the user’s Mac.

AI is also changing how companies organize work and gather training material. ClickUp rewards employees who build useful agents, measuring contributions through revenue acceleration, cost elimination and time savings, as well as executive-selected projects. Scaler cofounder Anshuman Singh and former AI product head Naman Bhalla launched Arcfront to turn professionals’ decision-making records into training data, practice environments and evaluations. Singh says four of five frontier AI labs became customers in Arcfront’s first year, but did not identify them; the company’s current ownership relationship with Scaler is unconfirmed.

Investor concerns extend beyond individual companies. One report says roughly 70% of technology-index value reflects expected future earnings growth, and recommends tracking adoption surveys, token use and prices, while assessing whether adoption produces durable earnings and returns. Another notes the Nasdaq Composite lost more than $4 trillion in market value in a week, sharpening questions about the consequences of an AI boom turning into a bust. AI-related concentration also affects retirement portfolios: technology represented 26.1% of the S&P 500 in 2020, compared with 21.4% in 2010, and tech stocks can be sensitive to rates, inflation and global events.

Enterprise spending is growing faster than oversight, according to NPI and Arcana Research’s 2027 planning report. AI accounts for a median 2.5% to 10% of IT budgets, rising to 14% to 20% at some mid-market companies and businesses with $2 billion to $10 billion in revenue. All 23 vendors assessed scored higher on execution than value, the report says, giving buyers room to negotiate. Arcana Research advises companies to improve visibility into usage and costs and reassess spending as upgraded models and open-source options become available. A video report on Anthropic’s public-market debut also highlights AI’s potential in healthcare, finance and transportation, alongside questions about employment and how society shares its benefits.

Key Takeaways

  • Third Point bought Nvidia, Microsoft and Meta shares in Q3 2026; the cited companies reported strong AI-related infrastructure, cloud and advertising growth.
  • Nvidia reported fiscal Q3 2027 revenue of $115.8 billion, up 103%, and forecast about $130 billion for Q4; AMD’s Q3 revenue rose 45% to $13.4 billion.
  • Anthropic says Claude Mythos can be moved to alternative chips in days, potentially increasing competition from Alphabet’s TPUs and AMD; TSMC could benefit from broader demand.
  • Broadcom forecasts AI semiconductor revenue of $115 billion in 2027 and $230 billion in 2028, supported by custom-chip work including TPUs developed with Alphabet.
  • On Semiconductor serves data centers, EVs and industrial markets; analysts expect free cash flow of $2 billion in 2027 and $2.3 billion in 2028, alongside risks from EV cycles and competition.
  • OpenAI says safeguards cut adaptive protected-reasoning extraction success for GPT-6.2 Astra and Sol from 54% to 0.2%.
  • ElevenLabs’ ElevenAI transcribes speech in more than 90 languages and creates meeting summaries, action items and follow-up drafts without adding a meeting bot.
  • ClickUp rewards employee-built AI agents for revenue gains, cost reductions, faster processes and executive-selected projects; Arcfront is building professional decision-making data and evaluations for AI labs.
  • NPI and Arcana Research estimate AI takes a median 2.5%–10% of IT spending, with some mid-market firms at 14%–20%; the report says governance and spending controls lag adoption.
  • AI investment risks include elevated earnings expectations, a technology-heavy S&P 500, and a reported Nasdaq loss of more than $4 trillion in market value during one week.

Dan Loeb buys four AI stocks, including Nvidia and Microsoft

Billionaire investor Dan Loeb’s Third Point hedge fund bought shares of Nvidia, Space Exploration Technologies, Microsoft, and Meta Platforms in the third quarter of 2026. Nvidia sells a full set of AI infrastructure products, and its revenue per gigawatt has grown from about $18 billion with Hopper to about $40 billion with Vera Rubin. SpaceX is expanding its AI infrastructure business and developing reusable rockets, while Microsoft’s Azure revenue grew 48% last quarter and its Copilot seats rose eightfold year over year. Meta uses AI to improve content recommendations and advertising, and its revenue grew 26% last quarter.

On Semiconductor offers a lower-risk way to invest in AI

The article argues that On Semiconductor may be a better long-term AI investment than Anthropic or Space Exploration Technologies. Its power and sensing chips serve electric vehicles, industrial equipment, data centers, and other markets, including products such as EV chargers and renewable energy systems. The company faces risks from the cyclical EV market, competition, and recent AI-related acquisitions, but its range of customers helps reduce its dependence on one area. Analysts expect it to produce $2 billion in free cash flow in 2027 and $2.3 billion in 2028, while its chips could support physical AI products such as robots, drones, and industrial machines.

Dan Loeb’s latest bets include Nvidia, SpaceX, Microsoft, and Meta

Third Point founder Dan Loeb bought shares of Nvidia, Space Exploration Technologies, Microsoft, and Meta Platforms in the third quarter of 2026. Nvidia’s AI infrastructure business generated about $40 billion in revenue per gigawatt with its Vera Rubin architecture, compared with about $18 billion with Hopper. SpaceX is working toward rapid reuse of its Starship rocket and is growing its AI infrastructure-as-a-service business. Microsoft reported 48% Azure revenue growth last quarter and an eightfold year-over-year increase in Copilot seats. Meta uses AI to improve recommendations and advertising, and its revenue rose 26% last quarter as its average price per ad increased 14% and ad impressions rose 10%.

AMD or Nvidia may be the better buy after the chip sell-off

Advanced Micro Devices shares fell 21% over the past month, while Nvidia shares dropped 40%, amid a semiconductor sell-off. A short-seller report made allegations about AMD, but Reuters said it could not independently verify them and AMD denied them. AMD’s third-quarter revenue grew 45% year over year to $13.4 billion, while data center revenue rose 97% to $8.5 billion. Nvidia reported third-quarter fiscal 2027 revenue of $115.8 billion, up 103%, and expects fourth-quarter revenue of about $130 billion. The article says Nvidia has stronger current revenue and profits but faces uncertainty as it searches for a CEO, while it considers both stocks attractive and Nvidia more compelling at its valuation.

Broadcom’s custom AI chips offer an alternative to Nvidia

The article presents Broadcom as an AI stock to consider while Nvidia searches for a new CEO after Jensen Huang stepped down. Broadcom helps companies design custom AI chips, including Tensor Processing Units developed with Alphabet that can have cost advantages over GPUs in some uses. Anthropic has built much of its computing infrastructure around these chips. Broadcom expects AI semiconductor revenue to reach $115 billion in 2027 and $230 billion in 2028, and its stock trades at 17 times next year’s earnings. The article also says Broadcom has well-known customers, a stable leadership team, and less risk tied to sales to China than Nvidia.

AI chip competition could pressure Nvidia and lift three rivals

As AI chips become more alike, Nvidia could lose market share and pricing power, which may shrink its profit margins. Anthropic said its Claude Mythos model can be shifted to alternative chips in days, compared with six to 12 months before, while Alphabet’s Tensor Processing Units and AMD’s GPUs are gaining ground. Taiwan Semiconductor Manufacturing could benefit from higher demand and a broader customer base. Alphabet may attract customers with lower-cost TPUs, while AMD could gain with lower prices, improving performance and its Helios AI rack system. Nvidia still has a hardware and software ecosystem and supply advantages in high-bandwidth memory and manufacturing.

Three AI chip rivals could gain as Nvidia faces more competition

AI models may soon run more easily on chips from different companies, which could weaken Nvidia’s market share, pricing power and margins. Anthropic said its Claude Mythos model could move to alternative chips in days, instead of taking six to 12 months and dozens of engineers. Taiwan Semiconductor Manufacturing could gain from greater demand and a wider mix of customers, while Alphabet could sell more of its lower-cost Tensor Processing Units. AMD may also benefit through lower prices, improving cost performance and its Helios AI rack system. Nvidia still has a connected hardware, software and networking system, along with supply advantages in high-bandwidth memory and manufacturing capacity.

Broadcom’s custom AI chips could make it a Nvidia alternative

Broadcom has become a major partner for companies designing custom AI chips, which can be better suited than Nvidia GPUs for specific workloads. Broadcom and Alphabet developed Tensor Processing Units, which can have a cost advantage over GPUs in some situations; Anthropic has built much of its computing infrastructure around TPUs. Broadcom said it expects AI semiconductor revenue to reach $115 billion in 2027, double that amount, or $230 billion, in 2028. The article argues that Broadcom could gain data center market share and points to its known customers, stable leadership and lower exposure to export risks than Nvidia. These growth figures are company projections, not guaranteed results.

OpenAI says GPT-6.2 Astra reasoning extraction fell to 0.2%

OpenAI said protections cut the success rate of attacks extracting protected reasoning from GPT-6.2 Astra from 54.0% to 0.2%. Extraction tied to the current request fell from 22.9% to 0.9%. The company uses classifiers to flag extraction attempts and block leaked reasoning in responses and tool-call arguments, and it ties protected reasoning to the conversation that produced it. OpenAI also investigates coordinated campaigns, bans violators and shares information with other providers. It said weekly enforcement actions in 2026 involved thousands to tens of thousands of accounts.

OpenAI reports stronger safeguards against reasoning extraction

On November 20, 2026, OpenAI said adaptive attacks extracting protected reasoning from GPT-6.2 Astra and GPT-6.2 Sol succeeded 0.2% of the time, down from 54.0% before the safeguards were strengthened. The evaluation combined results from six attacker models that adjusted their methods based on what they could observe. OpenAI said exact reasoning traces can transfer skills, including cyber skills, more effectively than final answers and summaries. Its safeguards check requests, assistant messages and tool-call arguments, and connect protected reasoning to the conversation history that created it. OpenAI also uses account-level signals to investigate coordinated extraction campaigns and said enforcement actions in 2026 ranged from thousands to tens of thousands of accounts each week.

ElevenLabs app turns speech into notes and meeting follow ups

ElevenLabs’ desktop app ElevenAI turns speech into text in more than 90 languages and works across apps. It can record in person and online meetings on a user’s computer without adding a meeting bot. After a meeting, it creates a summary, labels speakers, lists action items and drafts follow up messages. The article says audio is not stored and transcript history stays on the user’s Mac.

AI adoption rises as investors weigh future earnings

The article says artificial intelligence is driving major changes across the economy and may reshape entire industries. AI appears to have entered a period of rapid mainstream adoption, but expectations for future earnings are also rising. About 70% of technology index value reflects expected future earnings growth, leaving valuations at risk if those expectations fall. The article recommends tracking adoption surveys, token use and prices, and focusing on companies that turn adoption into lasting earnings and strong returns on invested capital.

Anthropic’s public debut highlights AI’s economic promise and risks

A video report discusses Anthropic’s debut on public markets and the growing role of AI in the U.S. economy. It says the S&P 500 had risen more than 20% that year and describes Anthropic as an AI startup valued at over $1 billion. The report notes AI’s potential to affect fields such as healthcare, finance and transportation. It also raises concerns about how AI could change jobs and how society will share its benefits.

ClickUp sets rewards for employees who build useful AI agents

ClickUp found that its most valuable AI agents were built by employees who understood where work processes often fail and could turn that expertise into clear rules. The company now uses a framework to value and reward employee skills used to create AI agents. It groups skills into four types: revenue acceleration, cost elimination, time compression and executive bounties. The first three are measured by added sales, reduced spending or faster processes, while executive bounties reward work on specific problems chosen by company leaders.

Scaler cofounder launches AI training data venture Arcfront

Scaler cofounder Anshuman Singh has launched Arcfront with Naman Bhalla, Scaler’s former head of AI products and an ex-Google engineer. The venture collects records of how professionals make decisions and turns them into training data, practice environments and tests for advanced AI models. Its work covers software engineering, cybersecurity and other fields, with sector-specific evaluations for areas including healthcare, legal services and finance. Arcfront operates under Evaratus Inc., which recruitment material identifies as formerly Scaler AI Labs, but its current ownership relationship with Scaler is unconfirmed. Singh said four of five frontier AI labs became customers within Arcfront’s first year, but he did not name them.

What an AI Market Bust Could Mean for Investors

The tech heavy Nasdaq Composite lost more than $4 trillion in market value this week. Friday’s market drop made the possible fallout from an AI boom turning into a bust a more urgent concern. The article considers what that downturn could mean, though the provided text does not include further details.

AI Stocks Add Volatility to 401(k) Plans

The S&P 500 has become more concentrated in technology stocks as AI has grown. Tech stocks made up 26.1% of the index in 2020, compared with 21.4% in 2010. The article says these stocks can react strongly to interest rates, inflation, and global events, adding volatility for investors. It suggests diversifying with non-tech stocks or considering index funds and ETFs.

Enterprise AI Budgets Grow Faster Than Governance

A report from NPI and Arcana Research says AI has become a major IT spending category, but buying practices, governance, and spending controls have not kept pace. The report, The State of Enterprise AI: 2027 Planning Edition, estimates that AI takes a median 2.5% to 10% of IT spending, rising to 14% to 20% at some mid-market companies and firms with $2 billion to $10 billion in revenue. All 23 vendors evaluated scored higher on execution than value, giving enterprises room to challenge prices and negotiate terms. Arcana Research managing partner Ian Kar says companies will need to review AI spending as upgraded models and open-source options emerge, while improving visibility into spending and use.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

Artificial Intelligence AI Infrastructure Nvidia Microsoft Azure Meta AMD Broadcom Alphabet TPUs Semiconductors OpenAI AI Safety ElevenLabs Enterprise AI AI Agents AI Training Data AI Spending AI Investment Risk Data Centers SpaceX Technology Stocks

Comments

Loading...