For UK retail investors, Donald Trump’s new ‘Super Intelligence’ initiative is less about changing what AI companies do today and more about removing one potential obstacle to the next phase of the AI build-out. Crucially, it seems clear that Washington wants the US AI industry to keep building rather than slow down.
On 29 September, Trump signed the White House Accord on Super Intelligence, alongside senior executives from Google, Meta, Nvidia, OpenAI, Anthropic and xAI. The voluntary agreement asks companies developing frontier AI models to introduce multiple layers of internal controls, independent auditing and board oversight. Trump simultaneously ordered US government agencies to use ‘Super Intelligence’ or ‘SI’ rather than ‘artificial intelligence’ in official documents.
The timing is significant. AI infrastructure spending is already accelerating dramatically, with Gartner forecasting $2.7 trillion of worldwide AI spending in 2026, up 49.5% year-on-year.
What exactly has been agreed?
The initiative is not a new AI law or regulatory regime. It is a voluntary commitment called the Joint Commitment on Frontier Responsibilities.
It establishes four broad layers of protection:
| Layer | What companies have agreed to do |
| 1. Internal controls | Monitor model capabilities and alignment during development and deployment |
| 2. Internal oversight | Have teams check that controls and monitoring systems actually work |
| 3. External audit | Use independent external auditors/evaluators |
| 4. Board oversight | Give an independent board committee responsibility for reviewing findings and ensuring problems are addressed |
The framework specifically references cybersecurity, biosecurity, chemical threats and preventing models from accessing or hacking technical systems unintentionally. Companies also agreed to meet regularly to develop common safety standards.
There is an important caveat for investors: the accord is voluntary and currently non-binding. The document leaves open the possibility that its principles could eventually be incorporated into legislation or regulation.
Trump described it as ‘morally binding’ and said it was ‘almost like a constitution’.
Which Big Tech companies signed?
The six headline signatories are:
- Alphabet/Google — Sundar Pichai
- Meta Platforms — Mark Zuckerberg
- Nvidia — Jensen Huang
- OpenAI — Greg Brockman
- Anthropic — Dario Amodei
- xAI — Elon Musk
Other major technology executives attended the White House meeting, including Microsoft CEO Satya Nadella, AMD CEO Lisa Su and representatives of Amazon and Palantir, but the six companies above are the named signatories to the accord.
That distinction matters: investors should not assume every company represented at the White House formally signed the document.
Why does this matter for AI stocks?
The most interesting investment implication is that safety is increasingly becoming part of the AI infrastructure story rather than simply a constraint on it.
Nvidia CEO Jensen Huang summed up the industry’s position:
‘There’s no conflict between innovation, technology and safety.’
Huang argued that the industry is developing new technologies both to increase AI capability and improve safety.
Meta’s Zuckerberg was similarly clear that the agreement was designed to give customers confidence that AI systems work as intended. But he also stressed that ‘this is a start’, rather than a finished regulatory framework.
Anthropic’s Amodei struck a more cautious tone, saying AI has ‘incredible benefits’ but that the technology also has ‘very real risks’ and that the mechanism for addressing those risks remains under discussion.
Anthropic IPO: what it means for UK investors in 2026
For investors, those comments reveal an important split: the industry broadly wants continued investment in AI, but there is increasing recognition that the ability to deploy increasingly powerful systems safely could become an economic requirement.
The numbers behind the next AI leg
The scale of the investment already under way is enormous.
TrendForce estimates that the nine largest cloud service providers could spend more than $886.7bn on capital expenditure in 2026, with the five major North American hyperscalers accounting for nearly 90%. AI server shipments are forecast to rise almost 31% in 2026.
Microsoft alone expects roughly $190bn of 2026 capital expenditure, while saying it remains capacity constrained through the year.
Recent estimates put 2026 capital spending at approximately:
| Company | 2026 capex guidance |
| Microsoft | ~$190bn |
| Alphabet | $195–205bn |
| Amazon | ~$220bn |
| Meta | $130–145bn |
| Four combined | ~$735–760bn |
These figures include some spending beyond AI, but AI infrastructure and data centres are the dominant driver of the increase.
Gartner’s latest forecast puts AI infrastructure spending at almost $1.98tn in 2026, making it by far the largest component of global AI spending.
That is why the safety accord matters beyond the headlines.
If better controls and independent verification make governments, businesses and consumers more comfortable deploying advanced AI, the potential addressable market expands rather than contracts.
What could be the next leg of the AI trade?
The first phase of the AI boom was dominated by chips and computing capacity.
Nvidia became the clearest beneficiary as hyperscalers raced to build GPU clusters. Semiconductor companies, networking suppliers, memory manufacturers, data centre operators and power infrastructure businesses subsequently joined the trade.
The next phase could increasingly move up the value chain.
1. Nvidia and the AI semiconductor ecosystem
For Nvidia, safety standards could ultimately reinforce demand for increasingly sophisticated computing.
More powerful models require more compute, while safety monitoring itself requires additional testing, evaluation and inference capacity.
That creates a potentially positive feedback loop:
More capable models → greater safety requirements → more testing/compute → more infrastructure demand
But Nvidia investors should also remember that expectations are already exceptionally high. The key question is therefore not whether AI spending is growing, but whether Nvidia can continue converting that spending into revenue and earnings growth.
2. Microsoft, Alphabet, Amazon and Meta
The hyperscalers are becoming increasingly important to the AI investment thesis because they control the infrastructure, cloud distribution and increasingly the models themselves.
Microsoft’s statement that it expects to remain capacity constrained through 2026 is particularly revealing.
For investors, the next question is gradually changing from:
‘Who is buying GPUs?’
to:
‘Who is generating sufficient revenue from AI to justify hundreds of billions of dollars of infrastructure investment?’
That is a much more demanding investment question.
3. AI software and agents
The potentially bigger prize is the transition from AI infrastructure to AI applications and agents.
Gartner has raised its forecast for AI application-development platforms to 39% growth in 2026, from 28% in its previous forecast, as businesses increasingly develop customised AI applications.
That could broaden the investment opportunity beyond the obvious chip and hyperscaler names.
Enterprise software, cybersecurity, data management, automation and professional services could increasingly participate in the AI spending cycle.
4. Energy, networking and data centres
There is another important consequence.
Trump simultaneously reiterated support for expanding America’s data centre capacity, despite growing concerns about electricity demand and local infrastructure.
That keeps the spotlight on the physical AI economy:
AI models → GPUs → networking → data centres → electricity → cooling → memory → construction.
For UK investors, this means the AI trade potentially extends well beyond US technology stocks.
The big risk: safety becomes regulation
There is also a clear bear case.
The agreement may be seen as insufficient by policymakers and the public because companies are largely responsible for defining and auditing their own controls. The accord does not currently create legally enforceable standards.
That leaves investors facing two opposing possibilities.
Bull case
- Voluntary safety standards increase public confidence.
- Governments remain supportive of AI infrastructure.
- Data centre construction accelerates.
- Enterprise adoption broadens.
- AI capex remains exceptionally strong.
- Semiconductor, networking, power and cloud suppliers continue benefiting.
- AI moves from infrastructure spending towards monetised applications.
Bear case
- Serious AI incidents increase political pressure.
- Voluntary standards prove inadequate.
- Governments introduce tougher regulation.
- AI deployment becomes more expensive and slower.
- Data centre projects face greater local opposition.
- Huge hyperscaler capex fails to generate sufficient returns.
- Investors begin questioning whether AI infrastructure spending has run ahead of monetisation.
That last point is particularly important. The next stage of the AI trade probably cannot rely indefinitely on the argument that companies are simply buying more computing power. Investors will increasingly want evidence that AI generates sustainable cash flows.
What UK retail investors should watch
The Trump accord therefore shouldn’t be interpreted simply as a new ‘AI safety’ investment theme.
It is better understood as a signal about the direction of the AI industry.
The US administration is explicitly trying to encourage rapid AI development while placing responsibility for safety largely on the companies themselves. At the same time, the largest technology companies are committing unprecedented sums to infrastructure.
That creates several indicators worth watching:
| Indicator | Why it matters |
| Hyperscaler capex | Shows whether AI infrastructure demand remains strong |
| AI revenue growth | Tests whether spending is becoming economically productive |
| GPU/ASIC demand | Measures underlying compute requirements |
| Data centre power availability | Potential constraint on further expansion |
| AI inference demand | Could become the next major source of semiconductor consumption |
| Enterprise AI adoption | Determines whether AI moves beyond experimentation |
| Safety incidents/regulation | Could alter deployment costs and timelines |
| AI margins/free cash flow | Critical test of whether the investment cycle is sustainable |
Sharesify investor verdict
The ‘Super Intelligence’ accord does not change the fundamental AI investment story overnight. What it does is provide another indication that Washington wants the US AI industry to keep building rather than slow down.
With Gartner forecasting $2.7tn of global AI spending in 2026 and hyperscalers committing hundreds of billions of dollars to infrastructure, the investment cycle remains exceptionally large.
For UK retail investors, the interesting question is therefore shifting from ‘Is AI a bubble?’ to something more specific: which companies can turn the enormous AI infrastructure investment cycle into durable revenue, earnings and cash flow?
That potentially makes the next phase of the AI trade broader — moving from Nvidia and the GPU supply chain towards hyperscalers, networking, memory, power infrastructure, cybersecurity and eventually the software and agents that monetise the computing build-out.
But the safety accord also highlights the central risk: the faster AI becomes more powerful, the more important trust, safety and regulation become to its commercial deployment.
Disclaimer: The author Steven Frazer has a personal interest in Scottish Mortgage, Nvidia.
You might also like:
