Headlines this week - Aug 24, 2025

A look at how capital is being deployed across future opportunities

C Gavilanes

Aug 24, 2025

This week in the future:

1 - The "Superintelligence" race is now also an existential battle for the consumer interface

The next frontier for AI is real-world data, and consumers apps / “super-apps” have plenty. Having exhausted public internet data, AI giants are now racing to secure real-world user data through apps and devices. This "elusive" data is now seen as the key ingredient needed to train the next generation of more capable AI models.

The browser is becoming a high-priority battleground. The FT reports that browsers are now a central element in the AI wars, as they can steer massive audiences to specific AI services. This explains why startups like Perplexity are investing so heavily in their own browser experiences to compete for users, and also why they’ve offered so much (see last week) to acquire Google Chrome.

A parallel war is being fought over AI-native devices. The battle extends to hardware, where analysts see Google "beating Apple on smartphone AI" with its deeply integrated Pixel phones (even if, for many reasons, not big shifts in market share are expected). At the same time, companies like OpenAI are reportedly developing their own AI-centric gadgets to create new ecosystems beyond the smartphone.

Now under pressure, Apple may turn to a rival to power its AI ambitions. Clearly behind its rivals in the race to offer AI-centric hardware, Apple could now be exploring a partnership with Google to power a revamped Siri with its Gemini models. This potential deal highlights the immense pressure Apple is under to catch up in the generative AI race.

2 - Meta rebuilds its AI strategy, embracing partnerships in a race to catch up with competitors

The company's AI division remains in a state of constant flux. Meta has again restructured its AI division, splitting it into four new groups amid what the NYTimes describes as "internal turmoil". This latest overhaul is aimed at accelerating the company's push toward “superintelligence”.

The company’s recent "acqui-hiring" moves have generated anti-trust concerns. The company's high-profile talent acquisitions are drawing scrutiny. An FT op-ed argues that these "billion-dollar 'acqui-hires'" are a way for Big Tech to neutralize nascent competitors without triggering antitrust oversight, a practice described as "bad for competition".

Meta is now freezing its aggressive AI hiring. Following a massive and costly spending spree, Meta is now freezing hiring for its AI division. This move comes as the company reorganizes its sprawling AI operation amid growing regulators’ concerns (see above) and investor scrutiny over costs.

It is also breaking taboos by partnering with arch-rival Google... In what looks like a significant strategic shift, Meta has signed a cloud deal with Google worth over $10bn. The company is moving beyond its own infrastructure to tap a key rival for the massive computing capacity needed for its AI models.

...and licensing technology from AI startup Midjourney. According to news this week, Meta will also license technology from image-generation startup Midjourney as its own in-house models lag rivals. The company’s management is describing this "all-of-the-above" approach as a sign of a new willingness to look outside for best-in-class technology.

3 - The Nvidia-China rollercoaster continues, creating new uncertainties

DeepSeek signals a shift toward homegrown Chinese chips. In a significant move, top Chinese AI lab DeepSeek has hinted that its next model will be compatible with domestic chips. This aligns with Beijing's push for technological self-reliance and reduces its dependency on Nvidia. This would also contradict previous rumors about local chips not being good enough.

The news has sparked a rally in Chinese semiconductor stocks. This announcement gave a major boost to China's homegrown chip industry, causing local semiconductor shares to surge. So investors are betting this signals a credible shift in demand from Nvidia to domestic suppliers.

Nvidia appears to be reacting with a new, more powerful China chip. In a sign that sales of the H20 may be struggling, Reuters reports that Nvidia is already working on a new, more powerful AI chip for China. This move suggests the company is adapting its strategy to compete more aggressively with the improving performance of local alternatives.

4 - The US government is taking a direct stake in Intel

A 10% stake marks a new era of industrial policy. The Trump administration has agreed to take a 10% stake in Intel, a move that, according to the WSJ, may be the first of many such deals. This direct investment is framed as a new, more transactional approach to the government's long-standing financial support for the chip industry.

The move is seen by some as a dangerous "nationalization”. A WSJ opinion piece argues that this move amounts to a de facto "nationalization of Intel". The author warns that government influence could lead to poor business decisions, such as preventing the company from making necessary but politically unpopular cuts.

Others see it as a pragmatic step in an "absurd world”. In contrast, other analysts frame the deal as a "sane" move that simply formalizes the government's existing influence. Given the geopolitical stakes, the article suggests that pre-emptively investing in a company "too important to fail" makes a certain kind of strategic sense.

5 - The controversy over AI's impact on the job market continues to grow

Companies are increasingly justifying mass layoffs with AI. TikTok is laying off hundreds of content moderators in the UK as it shifts to a greater reliance on AI. This is a very tangible confirmation of the prediction that AI is on the way to fully eliminate some / most / all (depends on who you ask) jobs.

A top AI insider has warned of the end of entire professional careers. A co-founder of Google's Generative AI team advised this week against pursuing law or medical degrees. He argues that AI is poised to "destroy" these careers before the current generation of students can even graduate. Yes, he was quickly dismissed by other experts (e.g. Pedro Domingos) but this just highlights the uncertainty / controversy.

However, in the real world, social pressure is already forcing companies to reverse course. Facing a potential political and social backlash, Australia’s largest bank has just reversed a plan to replace 45 customer service roles with AI. Obviously, the transition to AI in the workplace is not frictionless and can trigger significant public pushback, especially if it includes layoffs…

A deeper question: is full automation even desirable? Even if AI can perform a task, there may be inherent value in the human experience of learning and doing, as an opinion piece at the FT claims this week. The author questions the wisdom of outsourcing all cognitive effort to technology, suggesting we lose something important in the process.

For instance, some domains might always benefit from a "human touch”. A recent blog post uses the world of professional sports (tennis) to make a case for human intuition over pure data-driven decisions. The piece suggests that even in highly analytical fields, the "human touch” remains a crucial and valuable element that can't be fully replicated by machines. So, even if there will be a temptation to fully automate many jobs, second-order implications could make us reconsider the plan, at least in some cases.

6 - More voices suggesting that a bubble is emerging around the AI investment boom

Investors are beginning to question sky-high AI valuations. This week an FT opinion column asked if it is "time to sell our AI stocks," questioning whether current massive valuations are sustainable. A key argument is the growing concern that, as discussed here last week, the path to AI monetization, especially in B2B, remains highly uncertain.

This week we saw a correction in the stock market. Tech stocks were hit by a significant sell-off this week, driven by concerns over the future of the AI boom. The end of the week was more positive, but this market reaction clearly revealed that investor skepticism about the sustainability of the current AI hype is growing.

Meanwhile, the pace of major infrastructure deals is not slowing down. Despite the market jitters, massive investments continue. As an example, the announcement this week that JPMorgan and MUFG are close to a $22bn data center financing deal. This highlights the immense capital that is still flowing into the AI infrastructure build-out. The fact that substantial amounts are now coming from “neutral” Private Equity firms, and not just from Tech Giants’ cash pools, reinforces the idea of instability.

7 - A "Nuclear Renaissance" is underway, driven by AI and new geopolitical ambitions

Big Tech firms are signing deals to acquire nuclear power from next generation reactors. This week we learned that Google has signed a landmark deal to buy power from a Small Modular Reactor being built by Kairos Power in Tennessee. The FT claims the deal signals growing confidence from the AI industry in advanced nuclear to power its massive data centers.

Nuclear is becoming a key geo-strategic priority for Taiwan... Taiwan is now weighing a return to nuclear power, driven by the energy demands of its chip sector and fears of a Chinese blockade. This move would be part of a global revival of nuclear energy as governments work to ensure energy security.

...and for Sweden. Sweden is also reversing its four-decade-old policy of abandoning nuclear power. The country is now planning to build several new Small Modular Reactors (SMRs) using either American or British technology to ensure stable, climate-friendly energy.

New use cases are also emerging, including a new space race for lunar reactors. A new race is heating up between the US, China, and Russia to build nuclear reactors on the Moon. The article explains that nuclear power is ideal to sustain a human lunar presence, given its energy density (a key advantage because the cost of getting things into space scales by mass).

8 - The future for renewables looks bright, even without subsidies

Solar and wind are now efficient enough to thrive on their own. A WSJ article claimed this week that solar and wind power can now thrive without subsidies, as their costs have fallen so dramatically. The article notes this makes them a "deflationary force" that will continue to gain market share even under less favorable political conditions.

Even nuclear-heavy France is adapting to a solar-powered grid. The rise of solar is reshaping the entire European energy market, forcing even nuclear-dependent France to "flex" its nuclear output to maintain grid stability. This new operational reality shows how renewables are becoming a dominant and system-defining force in the energy mix.

9 - The business press is calling to prepare for the "looming quantum era"

For the FT, recent breakthroughs suggest the technology is arriving sooner than expected. An FT editorial this week argues that the world must prepare for the "looming quantum era". The piece notes that recent breakthroughs have fueled optimism that workable quantum computers are in sight, making it imperative to start preparing for their disruptive potential now.

10 - "AI clones" are coming, bringing new opportunities… and unavoidable risks

Influencers and politicians are now creating their own AI clones. Public figures are currently creating AI versions of themselves to make their communication more efficient. A Parmy Olson article about this, titled "The AI Clones Are Coming! But It’s Not All Bad," suggests this is becoming a “must have” new tool for managing a public persona.

These digital replicas are increasingly being used in advertising. The NYTimes details how one creator recently sold his likeness, and now his AI avatar is being used to sell supplements on TikTok. This highlights the emergence of a new, scalable form of influencer marketing powered by AI-generated deepfakes.

On the negative side, this trend is fueling a new wave of sophisticated scams .This technology also has a dark side. AI (with the “cloning” tools just discussed) is actually driving a sharp rise in "CEO impersonator scams". These sophisticated deepfakes are being used to create convincing fraudulent requests, leading to significant financial losses for companies.