2025 has seen a material decline in the fortunes of the greenback. A technical structure breakdown early in the year was followed by a breach of the 200-day moving average (MA) at the end of Q1. The index then entered correction territory, printing a three-year low at the end of Q2.

Since then, we have seen attempts to build a technical base, including a re-test of the end-of-June lows in mid-September. However, buying pressure has not been strong enough to push price back above the technically critical and psychologically important 100 level.

What the levels suggest from here

As things stand, the index remains more than 10% lower for 2025. On this technical view, the index may revisit the 96 area. However, technical levels can fail and outcomes depend on multiple factors.

US dollar index

The key question for 2026

The key question remains: are we likely to see further losses in the early part of next year and beyond, or will current support hold?

We cannot assess the US dollar in isolation and any outlook is shaped by internal and global factors, not least its relative strength versus other major currencies. Many of these drivers are interrelated, but four potential headwinds stand out for any US dollar recovery. Collectively, they may keep downside pressure in play.

Four headwinds for any US dollar recovery

1. The US dollar as a safe-haven trade

One scenario where US dollar support has historically been evident is during major global events, slowdowns and market shocks. However, the more muted response of the US dollar during risk-off episodes this year suggests a shift away from the historical norm, with fewer sustained US dollar rallies.

Instead, throughout 2025, some investors appeared to favour gold, and at other times, FX and even equities, rather than into the US dollar. If this change in behaviour persists through 2026, it could make recovery harder, even if global economic pressure builds over the year ahead.

2. US versus global trade

Trade policy is harder to measure objectively, and outcomes can be difficult to predict. That said, trade battles driven by tariffs on US imports are often viewed as an additional potential drag on the US dollar.

The impact may be twofold if additional strain is placed on the US economy through:

• a slowdown in global trade volumes as impacted countries seek alternative trade relationships, with supply chain distortions that may not favour US growth
  
• pressure on US corporate profit margins as tariffs lift costs for importers
  

3. Removal of quantitative tightening

The Fed formally halted its balance sheet reduction, quantitative tightening (QT), as of 1 December 2025, ending a program that shrank assets by roughly US$2.4 trillion since mid-2022.

Traditionally, ending QT is seen as marginally negative for the US dollar because it stops the withdrawal of liquidity, can ease global funding conditions, and may reduce the scarcity that can support dollar demand. Put simply, more dollars in the system can soften the currency’s support at the margin, although outcomes have varied historically and often depend on broader financial conditions.

4. Interest rate differential

Interest rate differential (IRD) is likely to be a primary driver of US dollar strength, or otherwise, in the months ahead. The latest FOMC meeting delivered the expected 0.25% cut, with attention on guidance for what may come next.

Even after a softer-than-expected CPI print, markets have been reluctant to price aggressive near-term easing. At the time of writing, less than a 20% chance of a January cut is priced in, and it may be March before we see the next move.

The Fed is balancing sticky inflation against a jobs market under pressure, with the headline rate back at levels last seen in 2012. The practical takeaway is that a more accommodative stance may add to downward pressure on the US dollar.

Current expectations imply around two rate cuts through 2026, with the potential for further easing beyond that, broadly consistent with the median projections shown in the chart below. These are forecasts rather than guarantees, and they can shift as economic data and policy guidance evolve.

The “Magnificent Seven” technology companies are expected to invest a combined $385 billion into AI by the end of 2025.

‍Microsoft is positioning itself as the platform leader. Nvidia dominates the underlying AI infra. Google leads in research. Meta is building open-source tech. Amazon – AI agents. Apple — on-device integration. And Tesla pioneering autonomous vehicles and robots.

With such enormous sums pouring into AI, is this a winner-take-all game?

Or will each of the Mag Seven be able to thrive in the AI future?

Microsoft: The AI Everywhere Strategy

Microsoft has made one of the biggest bets on AI out of the Mag Seven — adopting the philosophy that AI should be everywhere.

Through its deep partnership with OpenAI, of which it is a 49% shareholder, the company has integrated GPT-5 across its entire ecosystem.

‍Key initiatives:

• GPT-5 integration across consumer, enterprise, and developer tools through Microsoft 365 Copilot, GitHub Copilot, and Azure AI Foundry
• Azure AI Foundry for unified AI development platform with model router technology
• Copilot ecosystem spanning productivity, coding, and enterprise applications with real-time model selection
• $100 billion projected AI infrastructure spending for 2025

Microsoft’s centrepiece is Copilot, which can now detect whether a prompt requires advanced reasoning and route to GPT-5's deeper reasoning model.

This (theoretically) means high-quality AI outputs become invisible infrastructure rather than a skill users need to learn.

However, this all-in bet on OpenAI does come with some risks. It is putting all its eggs in OpenAI's basket, tying its future success to a single partnership.

Elon Musk warned that "OpenAI is going to eat Microsoft alive"[/caption]

Google: The Research Strategy

Google’s approach is to fund research to build the most intelligent models possible. This research-first strategy creates a pipeline from scientific discovery to commercial products — what it hopes will give it an edge in the AI race.

‍Key initiatives:

• Over 4 million developers building with Gemini 2.5 Pro and Flash
• Ironwood TPU offering 3,600 times better performance compared to Google’s first TPU
• AI search overviews reaching 2 billion monthly users across Google Search
• DeepMind breakthroughs: AlphaEvolve for algorithm discovery, Aeneas for ancient text interpretation, AlphaQubit for quantum error detection, and AI co-scientist systems

Google’s AI research branch, DeepMind, brings together two of the world's leading AI research labs — Google Brain and DeepMind — the former having invented the Transformer architecture that underpins almost all modern large language models.

The bet is that breakthrough research in areas like quantum computing, protein folding, and mathematical reasoning will translate into a competitive advantage for Google.

Today, we're introducing AlphaEarth Foundations from @GoogleDeepMind , an AI model that functions like a virtual satellite which helps scientists make informed decisions on critical issues like food security, deforestation, and water resources. AlphaEarth Foundations provides a… pic.twitter.com/L1rk2Z5DKk

— Google AI (@GoogleAI) July 30, 2025

Meta: The Open Source Strategy

Meta has made a somewhat contrarian bet in its approach to AI: giving away their tech for free. The company's Llama 4 models, including recently released Scout and Maverick, are the first natively multi-modal open-weight models available.

‍Key initiatives:

• Llama 4 Scout and Maverick - first open-weight natively multi-modal models
• AI Studio that enables the creation of hundreds of thousands of AI characters
• $65-72 billion projected AI infrastructure spending for 2025

This open-source strategy directly challenges the closed-source big players like GPT and Claude. By making AI models freely available, Meta is essentially commoditizing what competitors are trying to monetize. Meta's bet is that if AI models become commoditized, the real value will be in the infrastructure that sits on top. Meta's social platforms and massive user base give it a natural advantage if this eventuates.

Meta's recent quarter was also "the best example to date of AI having a tangible impact on revenue and earnings growth at scale," according to tech analyst Gene Munster.

However, it hasn’t been all smooth sailing for Meta. Their most anticipated release, Llama Behemoth, has all but been scrapped due to performance issues. And Meta is now rumored to be developing a closed-source Behemoth alternative, despite their open-source mantra.

Amazon: The AI Agent Strategy

Amazon’s strategy is to build the infrastructure for AI that can take actions — booking meetings, processing orders, managing workflows, and integrating with enterprise systems.

Rather than building the best AI model, Amazon has focused its efforts on becoming the platform where all AI models live.

‍Key initiatives:

• Amazon Bedrock offering 100+ foundation models from leading AI companies,  including OpenAI models.
• $100 million additional investment in AWS Generative AI Innovation Center for agentic AI development
• Amazon Bedrock AgentCore enabling deployment and scaling of AI agents with enterprise-grade security
• $118 billion projected AI infrastructure spending for 2025

The goal is to become the “orchestrator” that lets companies mix and match the best models for different tasks.

Amazon’s AgentCore will provide the underlying memory management, identity controls, and tool integration needed for these companies to deploy AI agents safely at scale.

This approach offers flexibility, but does carry some risks. Amazon is essentially positioning itself as the middleman for AI. If AI models become commoditized or if companies prefer direct relationships with AI providers, Amazon's systems could become redundant.

Nvidia: The Infra Strategy

Nvidia is the one selling the shovels for the AI gold rush. While others in the Mag Seven battle to build the best AI models and applications, Nvidia provides the fundamental computing infrastructure that makes all their efforts possible.

This hardware-first strategy means Nvidia wins regardless of which company ultimately dominates. As AI advances and models get larger, demand for Nvidia's chips only increases.

‍Key initiatives:

• Blackwell architecture achieving $11 billion in Q2 2025 revenue, the fastest product ramp in company history
• New chip roadmap: Blackwell Ultra (H2 2025), Vera Rubin (H2 2026), Rubin Ultra (H2 2027)
• Data center revenue reaching $35.6 billion in Q2, representing 91% of total company sales
• Manufacturing scale-up with 350 plants producing 1.5 million components for Blackwell chips

With an announced product roadmap of Blackwell Ultra (2025), Vera Rubin (2026), and Rubin Ultra (2027), Nvidia has created a system where the AI industry must continuously upgrade to Nvidia’s newest tech to stay competitive.

This also means that Nvidia, unlike the others in the Mag Seven, has almost no direct AI spending — it is the one selling, not buying.

However, Nvidia is not indestructible. The company recently halted its H20 chip production after the Chinese government effectively blocked the chip, which was intended as a workaround to U.S. export controls.

Apple: The On-Device Strategy

Apple's AI strategy is focused on privacy, integration, and user experience. Apple Intelligence, the AI system built into iOS, uses on-device processing and Private Cloud Compute to help ensure user data is protected when using AI.

‍Key initiatives:

• Apple Intelligence with multi-model on-device processing and Private Cloud Compute
• Enhanced Siri with natural language understanding and ChatGPT integration for complex queries
• Direct developer access to on-device foundation models, enabling offline AI capabilities
• $10-11 billion projected AI infrastructure spending for 2025

The drawback of this on-device approach is that it requires powerful hardware from the user's end.  Apple Intelligence can only run on devices with a minimum of 8GB RAM, creating a powerful upgrade cycle for Apple but excluding many existing users.

Tesla: The Robo Strategy

Tesla's AI strategy focuses on two moonshot applications: Full Self-Driving vehicles and humanoid robots.

This is the 'AI in the physical world' play. While others in the Mag Seven are focused on the digital side of AI, Tesla is building machines that use AI for physical operations.

‍Key initiatives:

• Plans for 5,000-10,000 Optimus robots in 2025, scaling to 50,000 in 2026
• Robotaxi service targeting availability to half the U.S. population by EOY 2025
• AI6 chip development with Samsung for unified training across vehicles, robots, and data centers
• $5 billion projected AI infrastructure spending for 2025

This play is exponentially harder to develop than digital AI, and the markets have reflected low confidence that Tesla can pull it off.

TSLA has been the worst-performing Mag Seven stock of 2025, down 18.37% in H1 2025.

However, if Tesla’s strategy is successful, it could be far more valuable than other AI plays. Robots and autonomous vehicles could perform actual labour worth trillions of dollars annually.

The $385 billion Question

The Mag Seven are starting to see real revenue come in from their AI investments. But they're pouring that money (and more) back into AI, betting that the boom is just getting started.

The platform players like Microsoft and Amazon are betting on becoming essential infrastructure. Nvidia’s play is to sell the underlying hardware to everyone. Google and Meta compete on capability and access. While Apple and Tesla target specific use cases.

The $385 billion question is which of the Magnificent Seven has bet the right way? Or will a new player rise and usurp the long-standing tech giants altogether?

‍You can access all Magnificent Seven stocks and thousands of other Share CFDs on GO Markets.

Over the past 3 months Nvidia has moved through ranges that some stocks don’t do in years, in some cases decades. Having lost over 35 per cent in the June to August sell off, it quickly bounced over 40 per cent in the preceding 20 days once it hit its August low as we build positions ahead of its results. These results delivered Nvidia style numbers with three figure growth on the sales, net profit and earnings lines but this did not appease the market, seeing it fall 22 per cent in a little over 8 days.

Which brings us to now – a new 16 per cent drive as Nivida reports it’s struggling to meet demands and that the AI revolution is translating faster than even it expected. This got us thinking – Where are we right “Now” in the AU players? Thus, it’s time to dive into the drivers for the Nvidia and Co.

AI players. Supersonic As mentioned, Nvidia’s results have been astonishing – and it still has time to do a US$50 billion buyback. It collected the award for becoming the world’s largest company in the shortest timeframe in the post-WWII era, think about that for one second – that’s faster than Amazon, Microsoft, Apple, Google, Shell, BP, ExxonMobil, TV players of the 60s and 70s.

So the question is how does it keep its speed and trajectory? Well that comes from what some are calling the ‘supersonic’ scalers. These are the players like Google, Amazon, Meta and Microsoft that are the users and providers of the AI revolution.

These are the players that have spent hundreds billions thus far on the third digital revolution. Let us once again put that into perspective, the amount of spending is (inflation adjusted) the same as what was spent during the 1960’s on mainframe computing and the 1990’s distribution of fibre-optics. So we have now seen that level of spending in AI the next step is ‘usage’ and that is the inflection point we find ourselves at.

Currently AI is mainly used to train foundational models and chatbots – which is fine but not long-term financially stable. It needs to move into things like productions – that is producing models for corporate clients that forecast, streamline and increase productivity. This is the ‘Grail’ This immediately raises the bigger question for now – can this Grail be achieved?

The Voices To answer that – let us present some arguments from some of AI’s largest “Voices” On the AI potential and the possibility of a profound and rapid technological revolution, Sam Altman, CEO of OpenAI, has claimed that AI represents the "biggest, best, and most important of all technology revolutions," and predicts that AI will become increasingly integrated into all aspects of life. This reflects a belief in AI's far-reaching influence over time. The never subtle McKinsey and Co. has projected that generative AI could eventually contribute up to $8 trillion to the global economy annually.

This figure underscores the massive economic potential of AI. The huge caveat: McKinsey's predictions are never real-world tested and inevitably fall flat in the market. This kind of money is what makes AI so attractive to players in Venture Capital.

For the VC watchers out there the one that is catching everyone’s attention is VC accelerator Y Combinator which is fully embracing the technology. Just to put Y Combinator into context, according to Jared Heyman’s Rebel Fund, if anyone had invested in every Y Combinator deal since 2005 (which would have been impossible just to let you know), the average annual return would have been 176%, even after accounting for dilution. Furthermore to the VC story - AI has accounted for over 40 per cent of new unicorns (startups valued at $1 billion or more) in the first half of 2024, and 60 per cent of the increase in VC-backed valuations.

So far in 2024, U.S. unicorn valuations have grown by $162 billion, largely driven by AI’s rapid expansion, according to Pitchbook data. So the Voices certainly believe it can be achieved. But is this a good thing?

The Good, the Bad and the Ugly AI is advancing at such a rapid pace that existing performance benchmarks, such as reading comprehension, image classification and advanced maths, are becoming outdated, necessitating the creation of new standards. This reflects the fast-moving nature of AI progress. For example, look at the success of AlphaFold, an AI-driven algorithm that accurately predicts protein structures.

Some see this as one of the most important achievements in AI’s short history and underscores AI’s transformative impact on science, particularly in fields like biology and healthcare. This is the Good. Then there is the 165-page paper titled "Situational Awareness" by Aschenbrenner which has predicted that by 2030, AI will achieve superintelligence and create a $1 trillion industry.

Also, a positive, but will consume 20 per cent of the U.S. power supply. These incredible predictions emphasise the enormous scale of AI and the impact it will have on industry, infrastructure and people. The latest Google study found that generative AI could significantly improve workforce productivity.

The study suggests that roughly 80 per cent of jobs could see at least 10 per cent of tasks completed twice as fast due to AI, which has implications for industries such as call centres, coding, and professional writing. This highlights AI's capacity to streamline tasks and enhance efficiency across various fields. However it also raises the massive concern around job security, job satisfaction and the socio-economic divide as the majority of those affected by AI ‘productivity’ are in mid to low scales.

Then we come to Elon Musk’s new AI startup, xAI, which raised $6 billion at a valuation of $24 billion this year. The company is planning to build the world’s largest supercomputer in Tennessee to support AI training and inference. This all sounds economically and financially exciting but it has a darker side.

These are the kinds of AI ventures that have seen ‘deep-fake’ creations. For example Musk himself shared a deep-fake video of Vice President Kamala Harris. This is the ugly side of AI and reflects the broader cultural and ethical issues surrounding AI-generated content.

Furthermore – we should always be forecasting both the good and the bad for investment opportunities. These issues are already attracting regulations and compliance responses. How impactful will these be?

And will it halt the AI driven share price appreciation? It is a very real and present issue. Where does this leave us?

The share price future of Nvidia and Co is clearly dependent on the longer-term achievement of the AI revolution. As shown, the supersonic players in technology and venture capital are betting big on AI, with predictions that it will reshape the global economy, industries, and even basic societal structures. However, there is still uncertainty about the exact timeline for these changes and how accurately the market is pricing in AI's potential.

The AI ecosystem is moving at breakneck speed, with new developments outpacing benchmarks and productivity gains reshaping jobs, but whether all these projections that range from trillion-dollar economies to superintelligence materialises remains to be seen. Thus – for now – Nvidia and Co’s recent roller-coaster trading looks set to continue.