After the SaaSpocalypse, investors are finding shelter in different Fortresses
"The new fortress”. ©2026. pitchhawk. All rights reserved.
TL; DR
The SaaSpocalypse didn't destroy the market. It clarified it. And what it clarified is that the most consequential investment opportunity of our generation is not sitting in a SaaS dashboard. It is sitting in a critical mineral. A miner. A power station. A step-down transformer. A copper cable. A data centre floor. A cooling system. A chip set. And most of all a power grid that was never built for what AI is now demanding of it. And all that’s driving intelligence-for-hire at never-before-seen adoption rates that make SaaS adoption curves look like pancakes.
At the same time, two civilisational forces are reshaping what investors are willing to back, and why. The first is the physical infrastructure super cycle that AI compute has triggered. The second is the agentic AI wave that is beginning to abstract away entire layers of the software stack, concentrating more and more demand into the physical layer underneath.
Businesses innovating in this new landscape face a problem few are talking about in the right terms. The investment thesis has changed. The investors have changed. And the language required to raise capital in this market is nothing like the MRR and Rule of 40 narrative that defined the “software is eating the world” epoch.
But are founders and builders listening to the signal?
pitchhawk is.
What the SaaSpocalypse actually corrected
For years the startup funding market rewarded narrative. Tech, TAM, team and a SaaS multiple that assumed the curve would keep bending upward forever.
Investors paid for the story. Then the story stopped compounding. Multiples compressed. Growth assumptions didn't hold. The capital that had chased narrative velocity found itself sitting in portfolios full of businesses that looked impressive on a slide but melted under the blowtorch of serious scrutiny.
The SaaSpocalypse was not a market failure. It was a market correction, and corrections clarify. They strip away decoration and reveal structure.
What the correction revealed was the difference between a business built solely on a recurring revenue narrative, and the new fortress business that thrives when the friction that made software a sticky business starts to dissolve.
And what that means is that when a sufficiently capable AI agent can approximate the function of a point solution, the question investors are asking is no longer "what is this revenue worth in perpetuity?" It is "how long does this revenue last?"
The game has changed, profoundly, for some. The capital markets didn't stop investing. They started investing differently.
The civilisational signal the market just sent
On 12 June 2026, SpaceX completed the largest IPO in history, raising $86 billion at a valuation of $1.77 trillion.¹
The prospectus stated a total addressable market of $28.5 trillion, described as the largest TAM in human history, of which approximately $25.6 trillion was attributed to artificial intelligence through its xAI subsidiary.¹ Morgan Stanley projected SpaceX revenue of $3.4 trillion by 2040.²
Whatever you think of the valuation, and serious analysts have raised legitimate questions about it, the signal is unambiguous. The capital markets have decided that the convergence of space infrastructure, AI compute, and global connectivity is not a sector bet. It is a civilisational bet. The 1492 moment. Extra-terrestrial GDP. A new economic geography that doesn't exist yet but that serious capital is already positioning to own. And the aspirational secret for how SpaceX is going to get there is on page 93 of the prospectus.
This is not the world the SaaS playbook was built for. And it is not the world that a free matching platform scoring your pitch 73 out of 100 was designed to navigate.
Chameleon code and the abstraction of the middle
At roughly the same moment the space economy was being repriced by the SpaceX IPO, a different civilisational force was playing out at ground level. No charge. Embedded in your browser. No training required. No inertia.
OpenClaw, the open-source agentic AI framework developed by Austrian programmer Peter Steinberger and published in late 2025, went viral globally in early 2026 with a speed that had no precedent. Within weeks it had accumulated more stars on GitHub than Linux, a foundational technology that took decades to reach that level of adoption.³ In China alone, daily AI token usage rose from 100 trillion at end-2025 to 140 trillion by March 2026, a 40% increase in under three months, driven directly by the shift from occasional chatbot use to persistent agentic workloads.⁴
It’s what we at pitchhawk call “chameleon code”. AI that doesn't just respond to instructions but executes tasks autonomously, continuously, across systems, in the background.
Claude Code. OpenClaw. The agentic wave which can spin up a business capability in hours that previously took months. It can reach scale at a speed no previous technology matched.
Monthly zoom: Salesforce reaches only 5× by month 24 — its hockey stick comes later. Anthropic hits 44× by month 17.
Both indexed to 1× at start. Sources: HockeyStickPrinciples.com · SemiAnalysis newsletter
Analysis and chart from the pitchhawk article: “The SaaSpocalypse was always coming — we just needed to see this chart” ©2026. pitchhawk. All rights reserved.
And Microsoft's Project Solara points to where this leads.
A wearable edge device, always on, always connected, that executes tasks on your behalf through Azure. The PC. The laptop. The mobile phone. The personal hardware stack that existed because intelligence had to live close to the human, that is the middle layer that agentic AI is beginning to delete.
Not overnight. But directionally, clearly, inevitably.
"Merging with the machine”. ©2026. pitchhawk. All rights reserved.
When the middle layer collapses, demand doesn't disappear. It concentrates. Every device that becomes redundant is replaced by permanent, continuous, always-on hyperscale cloud compute demand. The physical stack doesn't shrink. And it walks, hears, sees, and actions, moving from distributed and local to concentrated and industrial.
Follow the physical stack
From the human wearing a Solara badge to the intelligence running in Azure, let’s trace every physical layer the signal passes through. The wearable edge device. The 5G network or LEO satellite mesh carrying the signal. The fibre. The data centre receiving it. The step-down transformer delivering power to that data centre. The copper in that transformer. The power grid behind it. The generation capacity feeding the grid. The minerals creating the feed.
Every layer is physical. Every layer is constrained. Every layer is attracting capital at a scale the market has never seen.
Alphabet, Amazon, Meta and Microsoft are on track to invest up to $725 billion in AI infrastructure capital expenditure in 2026 alone.⁵ More than half of US data centres planned for 2026 have been delayed or cancelled, not for lack of money or land, but because the transformers needed to step down grid power and distribute it safely inside hyperscale facilities are in critically short supply.⁶
Lead times for high-voltage transformers have stretched from 12 months to 36 to 48 months.⁶ In Australia, data centre investment drove a 6.5% rise in private capital expenditure in the March quarter 2026, the seventh consecutive quarter of growth in the sector.⁷ Australian data centre energy demand is forecast to grow from 3.9 TWh in FY25 to 14 TWh by 2030.⁸
This is not a technology story. It is a physical infrastructure story. And it was underway long before any wearable device shipped. It’s civilisational in nature.
Why AI won't abstract this layer away
Software has abstracted away hardware before. The cloud killed the on-prem server. SaaS killed installed software. The smartphone killed the desktop. Each wave deleted a layer of physical infrastructure that seemed permanent.
The question worth asking is whether AI will do the same to the physical enabling infrastructure.
It won't. Not this layer. You cannot run AI without electricity. You cannot deliver electricity without copper. You cannot step down grid voltage with software. You cannot cool a hyperscale data centre with an algorithm. You cannot manufacture a transformer with a language model, no matter how large.
What AI will abstract away is everything above this layer. Software categories, process steps, human workflows, entire business models built on the old stack. OpenClaw is already doing this. Solara points to more of it coming. But the physical infrastructure underneath the intelligence gets more essential, not less, every time another layer above it disappears. The abstraction of the middle concentrates demand into the physical foundation. The foundation cannot be abstracted. It can only be built.
The investment thesis has changed, have you noticed?
Here’s the problem that few are talking about clearly.
The physical infrastructure super cycle has historically been private equity territory. Patient capital. Long duration. Profit first, bank loans, then private equity for growth or buyout. The language was yield, depreciation, asset base, regulated returns, profit history, long-term annuity style revenues, contracts, offtakes, infrastructure-grade cash flows. Venture capital didn't go near it. Family offices held infrastructure assets quietly. The return profile was stable and unglamorous.
That world has changed.
The same physical infrastructure is now being chased by VC funds that three years ago only backed SaaS companies. By family offices making civilisational thematic bets. By sovereign funds investing in geopolitically-reshored, sovereign capability. By superannuation funds that have never deployed capital into a rocket company before Gilmour Space, Rocket Labs, or SpaceX.
They are not investing in yield and in some cases, profits will be years away. What they are investing in is good old creative destruction that will drive the next $30 trillion in new GDP. The physical layer that enables AI to destroy the software layer above it, which in turn creates the next wave of demand for the physical layer below it. It’s a self-reinforcing super cycle and they can see it.
The result is a completely new hybrid investor sitting across from you and wondering how you are leveraging these tailwinds and dodging the headwinds. Can you hear them?
Former SaaS VCs who have learned enough infrastructure language to be dangerous. Family offices making generational bets on civilisational themes. Sovereign funds with a strategic agenda that has nothing to do with IRR, and more to do with offtakes. And none of them speak the same language as each other, or as the founder sitting across from them.
A white coat building a transformer manufacturer, a grid technology company, a liquid cooling system, or a power infrastructure business has always needed to speak private equity language. Asset base, capex recovery, regulated returns, long duration cash flows. Not anymore.
Now they also need to speak the language of the VC making a civilisational thematic bet. And the language of the sovereign fund protecting national strategic capability. And the language of the family office investing across a generation.
The innovator/builder who only speaks the old recurring revenue SaaS thesis is invisible to all of them.
And the sell-side machinery most founders are relying on to build the old investment thesis that was all bour MRR (the accelerator curriculums, the matching algorithms, and boot camps that cling to the old view) no longer works, because it was built for SaaS. That machinery cannot ask the right questions of a business where the asset is a grid connection agreement or a step-down substation. It sees a bitcoin mining operation and says it’s manufacturing digital rat poison, ridiculously missing the point that it’s now energising for compute, not for mining. It cannot score that out of 100. It cannot match it to a mandate-aligned angel on a platform.
This is where the business investability gap is most dangerous. And most invisible.
The fortress in the new landscape
The SaaSpocalypse clarified what the market now values. OpenClaw and the agentic AI wave showed how fast the software layer above the physical stack can change. The SpaceX IPO confirmed where some of the largest thematic bets are being placed, even at the massive IPO price. And the energy and transformer shortage showed that the physical layer underneath it all is the binding constraint on everything.
The fortresses that investors are now running to are being built from different materials than before. Copper. Power. Transformers. Data centres. Fibre. Grid capacity. Chips. Space infrastructure. The unglamorous, irreplaceable, impossibly constrained physical stack that makes renting intelligence possible.
And about that adoption chart in the middle of this note — which curve do you prefer?
The standard hasn't changed. The fortress still needs foundations that are solid. Walls that hold under the hardest questioning. A commercial engine that accelerates uphill. A moat with real physicality, not narrative. And an investment thesis that speaks the right language to the right capital, and there are now more languages to speak than ever before.
The business investability gap doesn't close by itself. And the fortress doesn't build itself. But it starts with one conversation between two human beings. No cohort. No captivity. No equity. No conflicts. No agenda but yours.
Ready to find out if your fortress is built for the market that exists now?
Mike 🖐
Innovation doesn't stall for lack of ideas.
It stalls in the gap between a great innovation and an investable business.
That gap never closed because nobody was incentivised to provide founders with an independent investor's lens.
pitchhawk is.
¹ CNBC, SpaceX IPO takeaways: SPCX closes at $161, jumping 19% after record debut, June 2026. cnbc.com
² Morgan Stanley via LinkedIn, SpaceX revenue could hit $3.4T by 2040, June 2026. linkedin.com
³ CNBC, China's Lobster buffet: China's tech firms feast on OpenClaw as companies race to deploy AI agents, March 2026. cnbc.com
⁴ China Briefing, China Agentic AI OpenClaw Boom: What the Surge Means for Businesses, April 2026. china-briefing.com
⁵ build.inc, AI Infrastructure Capex in 2026: What Hyperscaler Spending Means for Developers, May 2026. build.inc
⁶ Energy News Beat, More than half of the Data Centers may be delayed due to lack of transformers and electrical equipment, April 2026. energynewsbeat.co
⁷ Australian Bureau of Statistics, Spotlight — Data Centres in Economic Statistics, 2026. abs.gov.au
⁸ Modo Energy, Australia NEM 2026 Data Centre Demand Forecast, 2026. modoenergy.com
© pitchhawk, 2025-6. All rights reserved. You may not copy, reproduce or imitate our services, content, frameworks or intellectual property.