The New Frontier - Your weekly science and innovation update

Defence: recruitment targets and record order books underline sustained momentum

Defence continues to provide hard evidence of sustained growth. This week Thales announced plans to recruit 9,000 people worldwide. What is of note, is the emphasis on expanding its technology and engineering workforce. It intends to hire 800 new employees in the UK in 2026. The UK expansion includes major investments in Northern Ireland and an AI centre, alongside a strengthened partnership with Faculty AI. BAE Systems, meanwhile, reported a 12% rise in full‑year operating profit, and its order book grew by £5.8 billion to a record £83.6 billion. It explicitly framed the outlook as a “new era” of defence spending that will support growth for years. Private equity interest is also picking up, with a broader rise in funding into aerospace and defence and a growing sense that the sector’s momentum is being underpinned by higher defence budgets, elevated geopolitical tension, and Europe’s move to build more independent capabilities.  In the UK, it was reported that Downing Street is considering a significant increase in defence spending, with proposals to meet a 3% of GDP ambition by the end of the current parliament. Industry is waiting for the long-delayed Defence Investment Plan which will lay out spending plans in more detail.

For property markets, the important point is that defence growth is not just about procurement. It is also about hiring, R&D and engineering, which pulls through demand for the right kinds of space.

AI fundraising: the mega round cycle continues

Fundraising headlines suggest investor appetite for AI remains strong, especially for platforms with strategic relevance to the AI supply chain. OpenAI is reportedly close to finalising a $100 billion funding round, potentially the largest capital raise in history which would boost its valuation to more than $850 billion. Despite being the world’s most valuable startup, OpenAI is expected to continue posting losses until at least 2029, which underlines how long dated the market is willing to be when it believes the underlying platform is foundational.

Elsewhere, a three‑month‑old startup formed by one of Britain’s top AI researchers is reportedly in talks to raise $1 billion, and AI pioneer Fei‑Fei Li’s World Labs raised $1 billion to accelerate work on “spatial intelligence” models that can perceive and reason about the 3D world, not just text and images.  These mega rounds are a reminder that in AI, exceptional people create enormous value, often turning small teams into category defining companies. For real estate, it’s therefore worth tracking where the top AI researchers and builders choose to work and live, because they are disproportionately likely to found (or attract) the next high-growth cluster anchor.

Big Tech earnings: the sector is building AI like a utility

The standout theme from earnings season isn’t just strong performance it’s the willingness to compress margins now to lock in capacity for tomorrow, with capital pouring into data centre infrastructure. Microsoft reported $37.5 billion of capex in a single quarter, while Meta is guiding $115 billion–$135 billion of 2026 capex largely to scale AI infrastructure. Alphabet went further, guiding $175bn–$185bn of 2026 capex and warning that depreciation and energy costs will weigh on the P&L, while Amazon expects around $200 billion of capex (mostly AWS) because it is “monetising capacity as fast as we can install it.”

At the same time, strategy is converging around AI agents. Microsoft is the most explicit, likening agents to “the new apps” and positioning its stack (Foundry, Fabric and Microsoft 365) as the platform for building, deploying, governing, and securing agents at scale. Amazon is rolling out production tooling to make agents safe to deploy, and Alphabet is pushing AI Mode and standardising agentic commerce, while Meta’s “personal superintelligence” vision includes agentic shopping tools.

The third headline is that energy is becoming the silent driver behind the AI buildout. Tesla framed this directly, arguing that solar and batteries are key to powering AI data centres and stating an ambition to work toward 100 gigawatts per year of solar cell production integrated across the supply chain. Across the broader sector, the implication is that the “AI infrastructure race” is now inseparable from securing the inputs that make compute possible, with energy a high priority.  Companies are bringing energy knowledge in-house and building their own supply, in some cases acquiring whole companies along the way.

Beneath the common themes, there is real divergence in what each company believes the prize looks like. Tesla is framing its next chapter around autonomy and humanoid robotics (Robotaxi/Cybercab and Optimus), supported by heavy capex and vertically integrated investments in batteries, compute and potentially solar. Meta is directing Reality Labs investment increasingly toward AI glasses and wearables, while simultaneously scaling infrastructure for its AI roadmap. Amazon’s opportunity set remains unusually broad—AI, chips, low-Earth orbit satellites, quick commerce, and everyday essentials—while AWS remains the primary engine and the primary capex sink. Microsoft is executing a clearly articulated three-part AI strategy: cloud “token factory” infrastructure, the agent platform layer, and high‑value agentic experiences (Copilot across domains). Apple, meanwhile, is leaning into AI as a mass‑market feature set (Apple Intelligence) and highlighted its wider industrial strategy, including a previously announced commitment to invest $600 billion over four years in the US across advanced manufacturing, silicon engineering, and AI

Liverpool–Oxford: the playbook for innovation clusters is collaborate don’t compete

A useful case study landed this week in the form of a Liverpool–Oxford partnership. The University of Liverpool, the University of Oxford, Oxfordshire County Council, and the Liverpool City Region Combined Authority have signed a Memorandum of Understanding committing the two ecosystems to work together on research, innovation, and commercialisation.

The ambition is deliberately “end-to-end”: connecting Oxfordshire and the Liverpool City Region to create a coherent pathway from research and company creation through to scale-up, industrialisation and global market growth, while helping firms and IP stay rooted in the UK and attracting domestic and international investment. The partnership also explicitly links two of the UK’s primary national research and innovation campuses operated by STFC—Daresbury (including the Hartree Centre) and Harwell (home to Diamond Light Source).

Why does this matter beyond Liverpool and Oxford? Because it exemplifies a broader, pragmatic trend: clusters are increasingly joining forces to compete globally, rather than competing domestically for the same pool of talent, capital, and attention. Scale increasingly comes from network effects between places.

AI talent watch: mission is moving the market

One of the most revealing themes in the AI cycle is that money no longer explains AI talent movement on its own. The Verge’s Decoder podcast captured the mood: the industry is “rife with defections, FOMO, and radical mission statements.”  It has scanned the internet to uncover what makes top AI talent move. A strong motivating source is ideology and mission. Talent is moving where a company’s values align with theirs and where they feel they can make the most positive societal impact.

The clearest example is the departure of Mrinank Sharma, who led safeguards research at Anthropic and resigned with a warning that “the world is in peril,” describing pressures that make it hard for organisations to let values govern actions, before saying he planned to pursue writing and a poetry degree. Whatever you make of the letter, it underlines a real point: top AI people are increasingly willing to leave prestigious roles if they believe the mission is drifting, the operating model is changing, or their impact is better expressed elsewhere.

At the other end of the spectrum, the recruitment tactics are becoming intensely personal. OpenAI’s chief research officer has said Mark Zuckerberg “hand-delivered soup” to prospective hires during Meta’s recruiting push an anecdote that is funny precisely because it is revealing: when compensation is already extraordinary, companies try to signal belonging, attention, and mission.

If the AI talent war is increasingly about ideology, then space becomes part of the proof. For occupiers, the workplace is no longer just where work happens; it is where you demonstrate mission and values. That is why “showcase space” starts to matter: auditoria, event programmes, demo zones, learning spaces, and places that connect teams to the wider ecosystem.

UKRI makes AI a funded, four-year programme and the government continues its partnership with tech

UKRI has launched its first ever AI strategy. It commits £1.6 billion over the next four years, with an explicit focus on expanding doctoral and fellowship routes co-designed with business to build the talent pipeline. It also sets out six priorities that span the full system: advancing technology development, transforming research through AI, developing skills and talent, accelerating innovation for economic growth and societal benefit, championing responsible and trustworthy AI, and building world‑class AI data and infrastructure.

Alongside the headline strategy, there are two concrete “compute” proof points, previously announced. UKRI will deploy up to £137 million as part of DSIT’s AI for Science Strategy to back AI-enabled scientific discovery, and it is providing £36 million to update the University of Cambridge’s DAWN supercomputer.

Separately OpenAI and Microsoft formally joined the UK’s international coalition to safeguard the development of advanced AI. The government has stated that safety remains central to the UK’s AI strategy.

Other reads:

The Singapore Life Sciences Trade Accelerator will select up to 30 UK businesses based on readiness to scale sales in the region, opening an export pathway.

New MHRA data shows a record number of approved clinical investigations in 2025, with approvals rising 17% compared with 2024. Neurotechnology saw the strongest growth, with investigations doubling since 2024.

OKYO Pharma announced the pricing of a $20 million underwritten public offering, offering 10,815,000 ordinary shares at $1.85 each. The funds are intended for clinical development, specifically for their neuropathic corneal pain (NCP) program, and working capital.