Atlassian cut roughly ten percent of its workforce this week. The company was explicit about why: it is reallocating toward AI and enterprise sales. Not disguising the layoffs as belt-tightening, not blaming the economy. Just stating that the bar for what software headcount is supposed to do has moved, and a meaningful portion of the existing team does not clear it anymore.
That is an uncomfortable thing to say in a press release. It is also the most honest framing of what is happening across enterprise software right now. The companies that built the last decade of SaaS are renegotiating their cost structures around the assumption that AI handles more of the work. The ones that are slow to do it will be outrun by those who are not.
The rest of the week had a clear shape. Two massive checks went into AI infrastructure. Google closed the largest software acquisition in history. A UK regulator finally handed Revolut a banking license after three years of waiting. Anthropic sued the Defense Department. And the macro backdrop got more complicated, not simpler.
The Capital Stack Is Splitting in Two
Nscale, a European AI infrastructure company, announced a $2 billion Series C at a $14.6 billion valuation. The lead investors were Aker ASA, a Norwegian industrial conglomerate, and 8090 Industries. The board additions included Sheryl Sandberg, John Decker, and Nick Clegg.
That last detail is worth pausing on. Sandberg, Decker, and Clegg are not venture partners helping a startup with its Series B. They are senior operators who run large organizations and navigate government relationships at scale. When a company installs that kind of governance structure, it is preparing for something: large sovereign contracts, a path to public markets, or both. Nscale is not being built to stay private and scrappy.
The same week, AMI Labs, the company co-founded by Yann LeCun, raised $1.03 billion at a $3.5 billion pre-money valuation. At seed stage. LeCun is the chief AI scientist at Meta and one of the handful of researchers whose name alone credibly moves a cap table. AMI Labs is pursuing what it calls world models, meaning AI systems that build internal representations of how the physical environment works rather than just predicting the next token in a sequence. The idea has been around in AI research for years. What is new is that investors are willing to write ten-figure seed checks to fund the attempt.
These two rounds are not the same story. Nscale is industrial infrastructure. AMI Labs is frontier research. What they share is scale that would have been unimaginable at these stages three years ago. The effect on the broader ecosystem is real: the bar for what counts as “differentiated” at early stages has risen sharply. A compelling pitch on AI infrastructure or foundational research now has to explain why it is not simply going to be crowded out by a $2 billion check from a sovereign-backed fund.
The more useful angle for early-stage investors is not to compete in those categories directly. It is to find what grows in their shadow. Nscale-scale buildouts create demand for orchestration tools, reliability monitoring, GPU billing infrastructure, and security tooling for AI clusters. None of those require balance-sheet scale. All of them benefit from the capex wave.
What $32 Billion Buys You
Google completed its acquisition of Wiz this week for $32 billion in cash. The deal had been announced last year and cleared regulatory review in both the US and Europe. Wiz builds cloud security infrastructure, specifically tools that help companies understand and manage risk across their cloud environments. It is infrastructure that security teams use constantly and almost never discuss publicly.
The price is the largest software acquisition ever recorded. And the strategic logic is straightforward enough: hyperscalers are competing on cloud security posture as much as on compute pricing and developer tools. Owning Wiz gives Google Cloud a meaningful asset to sell into enterprises that are running workloads across multiple providers and need a unified security layer. It is also a signal to the rest of the security market that cloud control-plane tools, identity management, and data security remain highly acquirable categories. Wiz will not be the last deal in this direction.
For founders building in security, the practical implication is about distribution. Google, Microsoft, and Amazon are all buying rather than building in key security domains. That means companies with genuine technical differentiation can find a path to exit through M&A rather than IPO, which changes how you think about the capital required and the timeline to liquidity.
Revolut, Eventually
Revolut received a full UK banking license this week. The approval came from the Prudential Regulation Authority after the company spent three years in a provisional authorization period called “mobilisation,” which is the UK regulator’s version of an extended dress rehearsal before issuing a full license.
Revolut is now one of the largest financial technology companies in Europe by customer count. It has been operating in the UK throughout the mobilization period, but without the ability to offer protected deposit accounts, which are insured up to £85,000 per customer. That changes now. The company plans to migrate existing UK customers to the bank over the coming months and launch new lending products once it has the infrastructure in place.
The patience required to get here is the actual story. Revolut filed its initial application in 2021. The delay was not a sign that the company was in trouble. It was a sign that UK banking regulation is genuinely demanding, and that the supervisory process moved at its own pace regardless of Revolut’s commercial momentum. The lesson for fintech founders is the same one that keeps being learned: regulatory timelines are not a function of how fast you can grow. Teams that underestimate compliance as a multi-year operational commitment tend to find out the hard way.
The Government Customer Problem
Anthropic sued the US Defense Department this week, challenging a designation that labels the company a supply chain risk under a federal security review framework. The practical effect of that label, if it stands, is that contractors and suppliers working with the Defense Department could be restricted from using Anthropic’s products in their own workflows.
This is a different kind of policy risk than the one covered in last week’s issue, where OpenAI negotiated contract terms that included explicit prohibitions on certain uses. That was a company choosing its limits before signing. What happened to Anthropic is a designation being applied from outside, by a regulator, with potential downstream effects across an entire category of customers. Anthropic did not sign a defense contract and then face internal backlash. It was effectively placed on a restricted list while conducting normal commercial operations.
The suit is a reminder that “government as customer” for frontier AI does not operate the way enterprise SaaS sales do. The government can also decide, for reasons that may or may not be publicly explained, to restrict access to a vendor’s products across its entire supply chain. Any company selling software into defense-adjacent markets, critical infrastructure, or regulated government environments needs to model this as a live risk, not a theoretical one. The question is not just whether you can sell to the government. It is whether the government can effectively unsell you to everyone else.
The Macro Collision Coming This Week
February CPI data came in steady. Headline inflation rose 0.3 percent month over month and 2.4 percent year over year. Core, which strips out food and energy, rose 0.2 percent month over month and 2.5 percent year over year. In isolation, this is a soft-landing reading. The Federal Reserve’s March meeting begins Monday.
The complication is that markets are not operating in isolation. The conflict in Iran has introduced meaningful uncertainty into global energy supply. The Strait of Hormuz handles a significant share of the world’s crude oil shipments, and a Congressional Research Service report published this week described a prolonged disruption as a scenario “with no historical precedent.” Oil has already crossed $90 per barrel.
The practical effect on venture and late-stage private markets is not abstract. Energy-driven inflation surprises change rate expectations. Changed rate expectations affect discount rates. Higher discount rates compress the valuations of long-dated, illiquid assets, which is precisely what secondary interests in private companies are. Deals will still happen. But buyers will ask for more protection: preferred return structures, price adjustments tied to future events, conditions that keep some of the value at risk until liquidity actually materializes. The CPI data says the trend is fine. The Hormuz situation says the trend may not hold.
What the Rest of the Week Was About
Wonderful raised $150 million at a $2 billion valuation, led by Insight Partners. The company sells AI agents into enterprise environments, but the detail worth noting is the operating model: heavy integration work, localization, market-specific customization. That is not a software company in the traditional sense. It is closer to a services business with AI at the center. The distinction matters for margins and defensibility. An AI agent company that wins through deep implementation has a real moat, but it also has real cost. Investors underwriting these companies need to price the deployment cost, not just the product.
Tropic closed a $105 million Series C, co-led by Forbion Bioeconomy Fund and Corteva. Climate tech funding has had a complicated few years, and this round stands out because it is structured around near-term commercialization rather than long-term promise. The strategic partners are agribusiness incumbents. The customers exist. That is a different kind of climate story than the one investors burned money on during the last cycle.
Cryptio raised $45 million for what it describes as ERP infrastructure for digital assets, meaning accounting and reconciliation software for institutions that hold or transact in crypto and tokenized securities. The pitch is not speculative: as institutional adoption of digital assets grows, the compliance and audit requirements grow with it. Cryptio’s software becomes mandatory, not optional, which is the characteristic of infrastructure that tends to stick.
UHS agreed to acquire Talkspace, the digital mental health platform. The buyer is one of the largest hospital systems in the United States. The exit mechanism was a strategic acquisition, not an IPO. That is a pattern worth tracking: digital health exits increasingly come from established healthcare providers buying distribution and patient access, rather than from public markets that have been skeptical of the sector’s unit economics.
What This Week Means
The clearest pattern from the week is the one Atlassian announced plainly at the start of it. Software headcount is being renegotiated around what AI can do. This is not a 2026 story. It has been happening for two years. But it is accelerating, and the companies making the shift explicitly, rather than quietly, are telling you something about where management confidence has arrived.
The capital bifurcation is real and worth holding clearly. On one side: multi-billion checks into AI infrastructure and frontier research, often with sovereign or industrial investors, often with governance structures designed for public-market readiness. On the other: execution-heavy application companies, climate businesses with near-term revenue, and infrastructure for the compliance requirements that follow AI deployment. The gap between these two pools is widening. The middle, meaning generic SaaS with AI features and no differentiated distribution, is getting harder to fund and harder to defend.
The Anthropic/DoD situation and last week’s OpenAI/Pentagon story are not isolated incidents. They are the opening moves of a longer dispute between frontier AI companies and governments trying to figure out how to regulate, procure, and restrict access to the technology at the same time. Founders building for government-adjacent markets should watch this closely. The commercial opportunity is real. So is the risk that a policy designation, rather than a product failure, becomes the constraint.
The macro backdrop adds a layer of volatility to all of this that is specifically relevant to late-stage private markets. The CPI data is fine. The energy situation is not resolved. The Fed meets Monday. How those three things land together will affect pricing in secondaries and late-stage rounds for the next quarter. The underlying investment thesis does not change. The terms on individual deals will.
Notable Rounds and Transactions
Nscale Series C · $2B · $14.6B
AI infrastructure at industrial scale. Board additions (Sandberg, Decker, Clegg) signal public-market preparation.
AMI Labs Seed · $1.03B · $3.5B pre-money
Yann LeCun co-founded. Frontier research (world models) attracting growth-stage capital at seed. Raises the bar on what “differentiated AI research” must look like.
Wonderful Series B · $150M · $2B
AI agents sold into enterprise through heavy integration and localization. Execution risk lives in deployment, not the model.
Tropic Series C · $105M · Unspecified
Climate bioeconomy funded on near-term commercialization + strategic partners (Corteva). A different climate story than the last cycle.
Cryptio Series B · $45M · Unspecified
ERP and audit infrastructure for institutional digital assets. Compliance-mandatory software tends to stick.
Newtrace Equity round · $6.3M · ~$30M post
Electrolyser components for green hydrogen. Picks-and-shovels play as supply chains localize.
Google / Wiz Acquisition · $32B (cash) · —
Largest software acquisition on record. Cloud security is a “buy not build” category for hyperscalers. M&A exit path for security founders.
UHS / Talkspace Acquisition · Unspecified · —
Digital health exit via strategic acquirer, not IPO. Established healthcare systems buying distribution and care-continuum access.
Sources
Nscale Series C: https://www.nscale.com/press-releases/nscale-series-c
AMI Labs seed round (TechCrunch): https://techcrunch.com/2026/03/09/yann-lecuns-ami-labs-raises-1-03-billion-to-build-world-models/
Google completes Wiz acquisition: https://www.googlecloudpresscorner.com/2026-03-11-Google-Completes-Acquisition-of-Wiz
Revolut UK banking licence (LSE): https://www.lse.co.uk/news/revolut-gets-full-uk-banking-licence-after-years-long-wait-3k8y73d6x1zqz6t.html
Tropic Series C: https://tropic.bio/series-c-raise/
Cryptio Series B: https://www.businesswire.com/news/home/20260312303330/en/Cryptio-Raises-%2445m-Series-B-as-Digital-Assets-Move-Into-Regulated-Financial-Markets
Atlassian workforce update: https://www.atlassian.com/blog/announcements/atlassian-team-update-march-2026
Anthropic sues Defense Department (NPR / Iowa Public Radio): https://www.iowapublicradio.org/news-from-npr/2026-03-09/anthropic-sues-the-trump-administration-over-supply-chain-risk-label
US CPI February 2026 (BLS): https://www.bls.gov/news.release/cpi.nr0.htm
Iran conflict and Strait of Hormuz (USNI): https://news.usni.org/2026/03/13/report-to-congress-on-the-iran-conflict-and-strait-of-hormuz
Wonderful Series B (TechCrunch): https://techcrunch.com/2026/03/12/wonderful-raises-150m-series-b-at-2b-valuation/
Newtrace funding (Economic Times): https://m.economictimes.com/tech/funding/climate-tech-startup-newtrace-raises-6-3-million-from-hdfc-bank-mitsui-sumitomo-peak-xv-others/articleshow/129378820.cms
Gemini in Workspace (Google Blog): https://blog.google/products-and-platforms/products/workspace/gemini-workspace-updates-march-2026
UHS acquires Talkspace: https://uhs.com/news/universal-health-services-inc-to-acquire-talkspace-inc/
Meta acquires Moltbook (TechCrunch): https://techcrunch.com/2026/03/10/meta-acquired-moltbook-the-ai-agent-social-network-that-went-viral-because-of-fake-posts/
NASA Artemis II readiness: https://www.nasa.gov/blogs/missions/2026/03/12/artemis-ii-flight-readiness-polls-go-to-proceed-toward-april-launch/