Beyond the AI bubble: where technology is driving progress in 2025 

The year 2025 could prove decisive for artificial intelligence. With spending on AI data centres set to exceed $1.4 trillion by 2027, the industry faces mounting pressure to deliver on its promises. Model makers boast that artificial general intelligence will change every aspect of our society and economy—as soon as this year. But much uncertainty remains. Some researchers have cautioned that the rapid progress of recent years may be slowing; most businesses adopting AI continue to struggle to extract a return on investment. Meanwhile, infrastructure and energy bottlenecks threaten to squander further progress.

But we shouldn’t lose sight of the bigger picture. Even with existing capabilities, AI is transforming our world, and we've only begun to see its far-reaching effects ripple through society. This presents many risks and challenges, but much of this progress offers enormous promise for society—as AI sets about transforming science, accelerating discovery and advancing human knowledge.

And technological progress is not confined to AI. Advances are being made across a range of fields: breakthroughs in climate technology, weight-loss drugs that could transform millions of lives, and quantum computing nearing practical applications. Here are key developments to watch in 2025:

AI’s hype fades, but its impact deepens

The peak of AI hype has given way to more sober expectations. Firms are struggling to derive meaningful value from generative-AI, while reports late in the year suggest that advances in core capabilities could be plateauing. The year ahead will be critical for the big model-makers to see whether a trillion-dollar bet pays off, as they seek a path to profit with a slew of “agents” that automate tasks. Time will tell whether the benefits will justify the expense for many.

But AI’s broader impact is undeniable. Over the past year, AI whizzery has helped preserve indigenous languages, restore the voice of an ALS patient and reconstruct the epic of Gilgamesh, one of humanity’s oldest texts—bringing “extreme acceleration” to ancient scholarship. Several drugs designed with AI are expected to enter phase-three trials this year. AI is also enabling breakthroughs in protein design, climate modelling and brain research, predicting clinical outcomes and future brain states just from scans. Whatever its future capabilities, the potential of current AI models has barely been explored.

In 2025, watch for how AI is beginning to change science and innovation itself. Researchers hope AI can be used to automate much of the scientific process—generating hypotheses, designing experiments and writing papers. If realised, this could create a cycle of accelerated innovation. Indicators to watch include the rise of AI-enabled startups in areas like protein design and forecasting, and the creation of new evaluation benchmarks for AI in science, which will be essential for building trust and assessing progress.

The accelerationist agenda

Donald Trump’s return to the White House could mean loosening AI regulations to boost innovation and America’s competition with China. In some respects that may be a good thing for innovation. But such moves are likely to provoke alarm among those focused on AI safety (which may include Mr Trump’s wingman, Elon Musk, who has both backed safety measures and invested heavily in AI). Expect more talk of “defensive acceleration”—the idea that sometimes the best way to mitigate technological risks is to develop better technology, to tackle challenges from AI and cyber-security to nuclear threats.

Next year’s AI Action Summit, a successor to those in Bletchley Park and Seoul, will reflect shifting priorities in technology governance on the global stage. The summit will drop “safety” from its title, signalling a pivot towards addressing practical risks and leveraging AI for public benefit in areas such as education, health, security and environmental sustainability. Expect less focus on speculative fears and wishy-washy pronouncements on aligning governance frameworks, and more emphasis on driving tangible outcomes.

A reckoning with AI’s hidden costs

The coming year will sharpen attention on the physical demands of artificial intelligence. Generative-AI is straining energy grids, with data centres consuming as much as 30% of grid capacity in some parts of the US. To meet demand, Microsoft will buy up all the electricity from the revived Three Mile Island nuclear plant. Others are exploring small modular reactors to power new facilities, though scaling nuclear energy is slow and contentious. AI’s energy use is keeping coal plants running, as seen in Omaha, where a power station serving Google and Meta scrapped plans to phase out coal. Data centres also gobble up vast amounts of water for cooling.

Assessing these impacts is difficult due to a lack of transparency, though this may soon change. The European Union will require data centres to disclose their energy and water usage, a move that could clarify their environmental costs and prompt debates about trade-offs. Can better laws reduce emissions? How will utilities, tech firms and regulators manage resource constraints? These questions will become pressing as AI expands.

Deep tech’s decisive moment

While AI dominates headlines, other technological frontiers are advancing rapidly and merit attention. 

In 2025, we’ll continue to see a shift in focus for innovation investment. The era of quick-scaling digital platforms and consumer apps is giving way to a more fundamental focus on harnessing and scaling scientific and engineering breakthroughs in frontier fields such as synthetic biology, advanced materials and quantum computing, to address humanity’s most pressing challenges. These technologies require more upfront investment and longer development cycles but promise more transformative impact. The doubling of deep tech’s share of global venture capital and private-equity funding from 10% to 20% in the past decade signals this strategic pivot.

The confluence of urgent global challenges, maturing technologies, and increasing institutional support suggests we’re entering what might be deep tech’s decisive moment. Governments, too, are looking to boost their deep- tech ecosystem in efforts that marry up a vogue for industrial policy and security policy in a more uncertain world, a fear of overreliance on foreign technology, with sustainability. Success will require new models of collaboration between research institutions, firms and governments—moving beyond the traditional venture- capital playbook to build sustainable innovation ecosystems capable of nurturing long-term technological breakthroughs.

Synthetic biology’s reality check

The synthetic-biology sector, in which microbes are modified to create products ranging from medicines to materials, is undergoing a reality check akin to that of AI. The three largest firms in the field stumbled in 2024, with bankruptcy, liquidation and lay-offs. Investors, frustrated by unmet promises, have pushed firms to shift from grand visions to narrower, practical applications.

The industry’s future lies in specialisation and practical applications over moonshot promises. The shift is particularly evident in pharmaceuticals, where biological therapeutics now represent a quarter of new drug approvals in America and generate nearly half of global pharmaceutical revenues. Meanwhile, groundbreaking research continues to showcase the field’s transformative potential. Researchers at Penn State University have engineered bacteria to detect TNT in soil, a development that points to powerful environmental- monitoring capabilities. For 2025, expect synthetic biology to follow a “less hype, more substance” trajectory akin to that in AI.

GLP-1 drugs: a health revolution in the making

Another breakthrough area to watch is GLP-1-based weight-loss drugs, which could transform society and reshape industries as their take-up spreads and new benefits emerge.  

Semaglutide, the active ingredient in drugs like Ozempic, will go generic in 2025 in key markets such as Canada, India, and China, expanding global access. Clinical trials investigating long-term effects, such as muscle-mass loss, will be critical as these drugs move from medical to lifestyle use.

Current users of GLP-1 drugs already consume 8% less food on average, a trend with far-reaching implications for healthcare, food industries, and consumer behaviour. And these drugs are proving effective not just at curbing obesity, but cardiovascular and kidney disease as well, and are now being tested for their powers to tackle Alzheimer’s and curb addiction.

A climate-tech tipping point

The International Energy Agency reckons that in 2025 combined solar-photovoltaic generation and battery storage will become cheaper than coal-fired power in China and new gas-fired plants in America. This will mark a crucial tipping point for clean- energy adoption. Meanwhile, hard-to-decarbonise sectors are seeing progress in carbon capture, green hydrogen, and grid-scale storage technologies, including sodium-ion batteries and compressed gas.

As the initial AI hype fades, a broader technological transformation is under way. From AI-driven science to breakthroughs in health and clean energy, 2025 may be the year when speculation gives way to substance and these innovations begin to fulfil their potential.