Table of contents
Green Bubble
After the pandemic, there were many warning signs of a possible bursting of the green investment bubble.
At the beginning of the year, with Trump’s re-election, the trend was easy to predict. Already in his first election, there had been a symbolic decision in his fight against ecological transition, namely the United States’ withdrawal from the Paris Climate Agreement and its almost complete absence from the United Nations COPs during his presidency. This was followed by the famous pre-election statements of his second campaign, such as, ‘We want to end the Green New Deal scam’ (August 2024). Finally, the icing on the cake: the conservative and sovereignist action plan, Project 2025. We wrote about it here.
And so, in the first half of 2025, investment in renewable energy in the United States had fallen dramatically: down $20.5 billion, a 36% drop compared to the second half of 2024.
On the other side of the ocean, things were no different. The latest developments from the USA, which usually arrive in Europe with a slight delay, were arriving almost simultaneously this time.
In January, with its competitiveness compass, Europe placed a strong emphasis on artificial intelligence, launching “a broad strategy on new technology, which also includes AI factories to enable companies to develop models for supercomputers”.
In February 2025, yet another alarm bell rang: the case of hybrid cars. Until then, the EU Commission’s decisions on the Green Deal meant that in 10 years’ time it would not be possible to buy cars that were not electric. However, European Commission President Ursula von der Leyen sought the opinion of Mario Draghi, who responded with his report giving the go-ahead for the production of hybrid cars after 2035.
On 4 March 2025, Ursula von der Leyen presented ReArm Europe in a letter to European leaders ahead of the European Council.
On 22 April 2025, the Commission proposed a regulation to encourage defence-related investment by allowing Member States, on a voluntary basis, to transfer resources from cohesion policy funds to programmes such as the European Defence Fund (EDF) and the Act in Support of Ammunition Production (ASAP), and also providing for more flexible use of cohesion funds for dual-use infrastructure (i.e. useful for both civilian and military purposes), such as military mobility. And in the context of the mid-term review of cohesion policy (2021-2027), the possibility of better aligning cohesion fund investments with emerging strategic priorities, including defence and security, was included.
So funds were diverted from Green, AI, to Defence. However, artificial intelligence could still rest easy thanks to the concept of dual use, a perfect glue.
Subsequently, in June 2025, the European Union took the first step in what could later represent a complete reversal of the Green Deal: the Omnibus package.
AI bubble
So, to understand that we could finally be witnessing the deflation of the AI bubble, this summer there was research from MIT warning everyone: “95% of generative artificial intelligence pilot projects in companies fail”.
To go beyond individual cases and superficial statements, the Financial Times published an article that analyses in depth the financial results transcripts and official documents sent to the SEC by companies included in the S&P 500 index. Unlike press conferences or public statements, these documents are mandatory and must include a detailed list of the risks perceived by companies, information that rarely reaches the general public. The data collected paints an interesting picture: although the adoption of AI technologies is on the rise and the tone of the discourse remains generally optimistic, the benefits described often appear vague or unspecific. In contrast, concerns, particularly those related to cybersecurity, are presented clearly and concretely. Furthermore, compared to 2022, there has been a decline in the number of companies expressing openly positive views on these technologies.
BlackRock, the perfect “track”
In reality, all you need to do is follow the money. Taking BlackRock, the world’s largest investment company, as a compass, the map becomes clearer: in January 2025, it will exit the Climate Fund.
In March, Nvidia and xAI (Elon Musk’s company) joined the AI infrastructure partnership already launched in September 2024 by BlackRock, Microsoft, GIP, MGX, and the Global AI Infrastructure Investment Partnership (GAIIP), with an initial goal of raising $30 billion in private equity and a total potential (with debt) of up to $100 billion.
Finally, in May, BlackRock launched the iShares Defence Industrials Active ETF (IDEF), an actively managed thematic fund providing exposure to companies active in the defence, aerospace, advanced technology and global security sectors.
Therefore, if the AI bubble does not implode, it will not be due to its private or corporate use, but to its use in warfare.
Although the Financial Times shows that investments in AI are not generating revenue for companies that adopt it, on the other hand, budgets for AI-related defence and security in 2025 have grown and, at least in the US, will soon clearly translate into orders/revenue for AI companies. In the EU and NATO, the push is strong, while Europe seems to be slower, but the reason is that tenders must first pass through Member States, as required by law, before they can be converted into actual orders for companies.
The Department of Defence’s new budget request calls for major investments in artificial intelligence and autonomy: large and small underwater drones, swarms of flying drones, flying swarms for both the Navy and Air Force, and basic research on artificial intelligence. And in July, the US Department of Defence’s Chief Digital and Artificial Intelligence Office (CDAO) announced the awarding of contracts to leading US companies in the field of frontier artificial intelligence to accelerate the adoption by the Department of Defence (now called the Department of War, following Trump’s decision) of advanced artificial intelligence capabilities to address critical national security challenges. It then awarded framework agreements (worth $200 billion each) to OpenAI, Anthropic, Google and xAI to develop advanced AI capabilities. This growth in budget and contracts will therefore translate into actual revenue for AI companies.
The EU, too, with its EDF 2025 work programme, has allocated one billion euros through public tenders with priority given to artificial intelligence relating to cybersecurity and defensive autonomy. In this case, Member States need only adopt the operational contracts to see the future revenues of these companies.
It should also be noted that NATO has allocated an additional €1 billion through its NIF fund to deep tech companies in the defence and security sectors. However, this is in the form of equity rather than operational contracts.
So, at present, it seems that the ‘war/security’ demand is acting as a safety net, reducing the risk of the AI bubble imploding. (photo by Nahrizul Kadri on Unsplash)
ALL RIGHTS RESERVED ©