This Week in Startups: 5 Stories That Actually Matter
$510 billion raised in six months, a $1.75B AI power deal, and venture capital's most uncomfortable data point yet.
Quick note before we start. SambaNova’s $1 billion round got its own post this week, so it’s not on this list. What’s left is five deals that share a theme. Almost all of this week’s money went to helping people own their stack instead of renting it. Here’s what actually mattered.
1. Oratomic raised $300M to skip straight to fault-tolerant quantum
Pasadena-based Oratomic announced a $300 million Series A, co-led by ARCH Venture Partners, Spark Capital, and Khosla Ventures, per the company’s announcement and reporting from The Quantum Insider and Yahoo Finance. Vinod Khosla posted on X that it was his firm’s largest initial investment since OpenAI. The company says its neutral-atom approach could reach useful quantum computing with roughly 10,000 qubits instead of the millions long assumed, though that number is a company-affiliated research claim, not settled science.
Spence’s take: When Khosla name-checks his OpenAI bet, study the structure of the wager, not the odds.
Here’s that structure, through the lens of a convex bet, a model from my book. The downside is capped at the check: $300 million, a knowable number. The upside, if fault-tolerant quantum actually arrives this decade, runs close to uncapped, from drug discovery to breaking the encryption that secures the modern internet. You don’t make that bet because you’re sure it works. You make it because the payoff is so lopsided that being wrong costs you one round while being right rewrites an industry. That asymmetry is also why generalist funds that kept quantum at arm’s length for a decade suddenly showed up at the table.
2. Norm AI hit unicorn status attacking the billable hour
Norm AI raised a $120 million Series C at a $1.2 billion valuation led by Khosla Ventures, per the company and TechCrunch. It embeds law and regulation into AI agents and runs an affiliated AI-native law firm that, according to its own materials, prices work on outcomes rather than billable hours and serves clients representing more than $30 trillion in combined assets under management (a company-reported figure).
Spence’s take: The scariest competitor isn’t the one with a better product. It’s the one whose pricing model your incumbents can’t copy without blowing up their own P&L.
3. Prime Intellect raised $130M so companies can be their own AI lab
Prime Intellect closed a $130 million Series A at a reported $1 billion valuation, led by Radical Ventures with NVIDIA Ventures, Intel Capital, and Dell Technologies Capital participating, per the company and TechCrunch. The pitch is to give enterprises the tooling to train and run their own agents instead of renting frontier APIs. TechCrunch reports the company has reached a $100 million annualized revenue run rate (company-reported).
Spence’s take: “Rent versus own” is the oldest question in business. AI just made it expensive enough to matter again.
4. Ollama raised $65M with 14 employees
Ollama announced a $65 million Series B led by Theory Ventures, bringing total funding to $88 million, per the company’s announcement. The kicker isn’t the money, it’s the leverage: the company reports 8.9 million monthly developers and a presence in 85% of the Fortune 500, run by a team of 14.
Spence’s take: Fourteen people, most of the Fortune 500. That ratio is the whole AI thesis in one line.
5. 8090 Solutions raised $135M to build software with agents on a leash
8090 Solutions, co-founded and led by Chamath Palihapitiya, raised a $135 million Series A led by Salesforce Ventures, per Crunchbase. The company builds enterprise software using coordinated AI agents kept under human-led oversight, a bet that the enterprise buyer wants autonomy and a human hand on the wheel at the same time.
Spence’s take: “AI does the work, humans own the outcome” is going to be the enterprise default. Anyone still pretending it’s one or the other is selling to the wrong decade.
The thread
Quantum hardware, legal work, model training, local inference, enterprise software. Five different sectors, one instinct: own the layer that matters instead of renting it. When capital moves this consistently in one direction, it’s usually pricing in a belief about who holds the leverage next.
Next week’s deep-dive goes straight at that shift. We’re digging into why the most regulated buyers on earth are pulling AI inside their own walls, and what the on-prem move means for anyone building in a compliance-heavy market. It’s the quiet story under half the rounds above.——————————————————————————————————————
If this was useful, share it with someone who builds things. And if you want the full toolkit of 50 mental models, you can grab my book, Mental Models: How to Think, Act, and Win, right now.
——————————————————————————————————————
Next week’s deep-dive:
Next week, I'm picking one company from this list and going deep: total funding, market size, business model, the bull case, and what could go wrong. Subscribe so you don't miss it.
——————————————————————————————————————