Davos, big tech talks about the future of AI

In Davos, while Donald Trump catalyzed political attention, the artificial intelligence was the most recurring dossier of World Economic Forum. In panels and conversations between companies and investors, i CEO they insisted on three issues: energy, data center infrastructure and economic returns on investments. The dominant feeling among the managers of the big tech is that AI is now a thing industrial and geopolitical and no longer just technological.

A clear message came from Jensen HuangCEO of Nvidiain a public conversation with Larry Fink (BlackRock). Huang rejected the idea that AI is a bubble, while acknowledging that it “looks like one,” arguing that the game is played on infrastructure: Investments in the order of “trillions of dollars” are needed to build the physical base that makes development possible. To explain it, he used the image of a multi-level supply chain, a “cake” that starts from energy, passes through chip e data centerand gets to models and applications. The point, according to Huang, is that the bottleneck today concerns the material capacity to power and cool the systems.

The list is simple: electrical powernetwork, systems, space, logistics. There calculation questionhe adds, is driving investments in infrastructure and, as a result, a new demand for technical work with profiles more linked to plant engineering, maintenance and commissioning of data center. Huang also drew attention to the risk that theEurope remains behind without an acceleration on energy capacity and data centers.

Larry Fink he framed the transformation from the point of view financialarguing that AI will continue to attract enormous capital. The theme of concentration: today few large platforms control critical infrastructureswith potential effects on competition and distribution of benefits. Jamie DimonCEO of JPMorgan, warned that AI “could go too fast for society” and cause “civil unrest” if governments and businesses do not accompany the workers who will be shocked by the change.

For Dimon the benefits will be significant (productivitymedical research), but adoption must be managed over time, with support programs and redevelopment. He also said JPMorgan could have fewer employees in five years with the introduction of AI. He also focused on the energy issue Andy JassyCEO of Amazon. He recognized one structural lack of electrical capacityinsufficient to support data center expansion. Amazon is evaluating direct investments in new energy sourcesincluding next-generation nuclear options. Satya Nadellanumber one of Microsoftshifted the discussion to “social license” of AI: if systems consume large amounts of energy without tangible benefits in health, education and productivity, they risk losing public consensus. Nadella rejected the idea of ​​a speculative bubble: the key step is the ability of companies to transform AI into operational tools.

Demis Hassabis and the AGI border

On the research laboratory front, one of the most awaited entries was that of Demis Hassabis (Google DeepMind) who participated in several sessions dedicated to the path towardsAGI (artificial general intelligence). Hassabis said that Chinese companies were not that far from the “Western technological frontier” but that they were still far from making a real breakthrough in terms of innovation. According to DeepMind’s guide, the main challenge now concernsreliability and the controllability of systemsin addition to their usefulness in scientific fields such as pharmaceutical research e climate simulation. These are, in fact, the two sectors in which we will probably see the first concrete and tangible effects of the impact of AI on our lives.

The growth paradox and social decoupling

Dario AmodeiCEO of Anthropicpresented in Davos a macroeconomic analysis based on a statistical paradox: global GDP growth of between 5% and 10% accompanied by an unemployment rate that could reach 10%. Amodei warned that the extent of the ongoing transformation is not yet fully understood, defining the scenario as a shift of such dimensions as to require a direct intervention by governments in the management of workforce.

According to Amodei the absence of a governance efficace would lead to a very significant social “decoupling”: an elite of around 10 million people, mainly concentrated in Silicon Valley, could benefit from individual growth rates of up to 50%, while the rest of the population would suffer the negative effects ofautomation. On the industrial level, the CEO of Anthropic drew a clear line between companies led by scientists and those run by leaders trained in the social media sector. For him, the scientific approach guarantees a greater intake of responsibility for the effects of technology. Anthropic, Amodei reiterated, will maintain its focus on enterprise sector to avoid the typical distortions of platforms oriented towards mass consumption.

Alex KarpCEO of Palantirstated that artificial intelligence will lead to the replacement of a large number of jobs in the humanities sector. Karp predicted that the job market will shift toward valorizing “outlier aptitude” and professional trainingpredicting that the demand for specialized technicians will exceed that of profiles with general university education. Karp also linked the efficiency of AI to a reduction in the need for mass immigration, arguing that automation will be able to cover domestic employment needs through vocational redevelopment of citizens. On a strategic level, the CEO reiterated Palantir’s centrality in defense and intelligencedefining AI not only as a productivity factor, but as ancritical infrastructure for national sovereignty.

Demand for computing, hardware, and investor skepticism

It was less visible in the media but central on an industrial level CoreWeavea company specializing in “computing power rental”. CEO Michael Intrator said the demand for GPU (high performance chips) remains very high, but i infrastructural constraints they are slowing down the expansion. The construction times of the data center and energy availability are the main factors of uncertainty. Intrator also spoke of a growing skepticism among investors: the phase of indiscriminate growth is giving way to a more rigid selection of sustainable projects. In other words, it is increasingly necessary to demonstrate a credible business model for AI. OpenAIhowever, maintained an institutional profile in Davos.

Chris Lehane, head of global relations, confirmed the company’s expansion beyond the boundaries of software, officially setting the window for the presentation of the first in the second half of 2026 hardware device created in collaboration with Jony Ive. The initiative has been described as a pillar of the annual roadmap, aiming for more natural human-AI interaction. The positioning in Davos was focused on “systemic responsibility“, with the announcement of new protocols for the safety of minors. OpenAI thus aims to consolidate its image as a reliable actor with governments. The maneuver also responds to the growing competitive pressure: the emergence of new efficiency standards from China forces the Californian company to accelerate on the hardware and production front infrastructural partnership.

Robotics, energy and AGI estimates

The intervention of Elon Musk outlined a convergence between artificial intelligence, robotics e energy infrastructureidentifying (again) in the availability of electricity the critical factor for data center expansion. Musk has taken issue with US tariffs on solar technology, calling them an economic hindrance. On the autonomous mobility front, Musk announced the imminent regulatory green light for the systems FSD (Full Self-Driving) in Europe and China, a preparatory step for the sui strategy robotaxi.

As for the humanoid roboticsthe roadmap of Optimus expects first operational use in factories by the end of 2026, with the aim of starting commercialization and complex industrial use by the end of 2027. Musk finally updated his estimates on theAGIpredicting the overtaking of individual human intelligence between the end of 2026 and 2027, and setting 2031 as the target for overtaking collective humanity. In addition to the panels dedicated to big tech, the Davos program hosted many parallel sessions, always dedicated to AI, with a recurring focus on Work, governance, cybersicurezza e technological sovereignty. The 2026 edition of the Davos Forum has therefore outlined an industrial sector focused on the resolution of infrastructural and energy limits necessary to support the current investment cycle. Despite warnings and calls for responsible management, the prevailing tone was operational optimismwhich finds a summary (almost involuntary) precisely in the words used by Elon Musk in greeting the audience: “In general, in life, I believe that it is better to err on the side of optimism and make mistakes rather than pessimism and being right”. An approach that, at Davos and beyond, will also continue to be applied to artificial intelligence.