Funding Flows to the Picks and Shovels of AI and Energy

The funding stories logged on this beat this week share one thread: investors are putting money into the physical layer beneath artificial intelligence, not into models or applications. Materials, compute capacity and electricity are being financed as the scarce inputs of the current buildout. The pattern matters for US technology companies because each of these raises is aimed at a constraint they will eventually pay for, directly or indirectly.

Three Rounds, One Direction of Travel

The largest of the three is Verda Cloud Oy, the Helsinki-based neocloud operator formerly known as DataCrunch, which announced $189 million in early-stage funding to expand its artificial intelligence cloud infrastructure, from AI compute capacity to platform services, as SiliconANGLE reported. Emergence Capital led the Series B, with participation from MUFG Innovation Partners, Supermicro, Varma Mutual Pension Insurance Company, Lifeline Ventures and others.

Smaller in headline size but similar in logic is Morphotonics, which raised €40 million to push its display technology into data centers, per TechCrunch. The investor group includes 3M Ventures, Innovation Industries, BOM and Invest-NL. A deeptech company with display roots is being financed specifically to serve data center demand.

The third is not a single round but a portfolio: Breakthrough Energy, backed by Bill Gates, has placed bets on 21 startups it expects to shape the future of energy, and its investors argue that the coming wave of electrification will make data center energy demands look modest by comparison, as TechCrunch reported.

The common denominator is that none of these are software companies in the conventional sense. They sell capacity, materials or generation. That is where the money is moving.

Why the Money Is Moving Down the Stack

AI spending has been concentrated at the top of the stack: chips and model development. Those layers are capital-intensive but also crowded, and their output is measured in benchmarks that change quickly. The layer beneath them is different. It is capital-intensive in a slower, more durable way. A data center, once built, produces revenue for years. A materials process, once qualified by a customer, is hard to displace.

Verda's raise illustrates the capacity side. The company is not selling a model; it is selling access to compute. It describes its expansion as running from AI compute capacity to platform services, which means the funding is aimed at both hardware footprint and the software layer that sits on top. That sequencing is a bet that demand for rented compute persists even as the largest buyers build their own.

Morphotonics illustrates the materials side. Its technology was developed for display manufacturing. Moving it into data centers means the company believes its process has an application in what goes inside or around the compute itself. The presence of 3M Ventures among the investors is notable: a large industrial materials company is putting money into a smaller deeptech firm with a data center thesis. That is a strategic bet on a supplier relationship as much as a financial one.

Breakthrough Energy illustrates the electricity side. The fund's framing, as reported by TechCrunch, is that electrification demand will dwarf what data centers consume today. That is a statement about the scale of the problem the portfolio is being assembled to address. It also explains why a fund with an energy mandate would be stacking a portfolio of 21 companies at once rather than concentrating on a handful. Energy is a systems problem, and systems bets are made across many positions.

The US Read-Through

The immediate question for US technology companies is cost. If compute capacity is being financed in Europe, as Verda's Helsinki base suggests, then US buyers of compute are competing for capacity with a better-capitalized European supplier base. That does not guarantee lower prices for US customers, but it does mean the supply side is broadening beyond the handful of US hyperscalers and their immediate partners.

There is also a supply chain implication. Morphotonics is a deeptech company whose investors include a Dutch regional development fund, a national investment vehicle and an American industrial giant. If its process becomes relevant to data center components, part of that supply chain may be developed outside the United States. US equipment makers and their customers should expect to see new entrants in component categories that have been stable for years.

The Breakthrough Energy portfolio points to a different US exposure: electricity prices. If the fund's thesis about electrification demand is broadly correct, the pressure on grid capacity and power pricing will not be confined to the regions where its portfolio companies operate. Large US data center markets already compete with residential and industrial users for power. A wave of new investment aimed at generation and grid capacity is a response to that competition, and it will take years to show up in supply.

Finally, there is a capital-markets read-through. All three stories involve investors taking positions in infrastructure with long payback periods. That is a different risk profile from software investing, and it suggests the funding environment for AI-adjacent companies is bifurcating: model and application companies compete for growth capital on short timelines, while the picks-and-shovels layer attracts patient capital with strategic co-investors. For US founders building in the physical layer, that is a friendlier market than for those building another application layer.

Where the Risk Sits

That said, patient capital still expects returns. Verda is raising at an early stage, and capacity businesses are vulnerable to oversupply if demand expectations are too aggressive. Compute capacity added today can become stranded if utilization falls, and neocloud operators have less financial cushion than hyperscalers.

Morphotonics is a deeptech company with a materials process. Its bet depends on qualification by data center customers, which is slow, technical and often binary. A €40 million round funds a runway, not a guaranteed commercial position.

Breakthrough Energy's 21 startups face the longest timelines of all. Energy infrastructure is regulated, permitting-dependent and geographically constrained. A portfolio approach spreads the risk, but it also means most positions will not produce near-term returns. The fund's own framing, that data center demand will look quaint by comparison, is a long-horizon claim, and long-horizon claims are the hardest to underwrite.

What to Watch

Three things will show whether this pattern holds. First, whether Verda converts its $189 million into named capacity and platform services, which would confirm demand for third-party compute beyond the largest US buyers. Second, whether Morphotonics announces data center customers rather than data center intentions, which would confirm that display-derived materials have a real route into the buildout. Third, whether Breakthrough Energy's portfolio companies announce commercial deployments, which would begin to test the fund's electrification thesis against actual grid and generation economics.

Until then, the signal from this week's funding is directional rather than decisive. Capital is moving down the stack. The question is whether the physical layer can absorb it at the pace the money implies.

More on this beat: Companies on TechManNews.

#funding#AI infrastructure#data centers#energy#deeptech#venture capital