Sharon AI’s $1.6B check puts compute back at the center - June 30, 2026

A $1.6 billion 'strategic financing' for a listed Australian neocloud - and not one product milestone in the press release to show for it. If you're just joining: AI infrastructure has been the gravity well on this beat. Baseten raised a $1.

5 billion Series F at up to a $13 billion valuation and Runpod followed with a $100 million Series A at a $1 billion valuation. Same basic shift: enterprise AI is moving from model demos into production and that's where inference capacity and developer infrastructure become budget lines not science projects. This is Startup Fundraising. Today - a recursive $200 million, a space-economy seed, and a neocloud that wants $1.

6 billion to build something it won't name. Sharon AI's up first. We're staying on AI inference infrastructure mega-rounds - follow the show and you won't miss what comes next. This one's from Sharon AI: SharonAI Holdings Inc.

(NASDAQ:SHAZ) and its subsidiaries (“Sharon AI” or “the Company”) a leading Australian Neocloud today announced the closing of its previously announced oversubscribed US$1.6 billion private placement financing (the “Transaction”). Let's be precise here. Sharon AI - ticker SHAZ on NASDAQ - closed $1.

6 billion, but this is a private placement into a public company, not a venture round. The press release calls it 'strategic financing.' I'd call it a structured debt-equity hybrid. Split the tape: roughly $900 million in equity and pre-funded warrants, plus $700 million in 4.

75% convertible notes due 2032. The convert lets them lever up without taking the dilution hit today. And the anchors? Situational Awareness and Oaktree - credit-oriented money, not your classic growth fund.

Right so a 'leading Australian Neocloud' raises one-point-six billion and the one thing they'll tell you it's for is buying up to 40,000 Grace Blackwell GPUs from NVIDIA. I want a business plan; what they gave us is a purchase order with a press release stapled to it. Same AI-infrastructure run, just a fatter check than anything we've seen. So who's the customer?

Forty thousand GB300s in Australia need sovereign contracts or hyperscaler offtake signed before the racks ship. Nobody disclosed a single one. And notice - no single lead setting terms. 'Anchored by' two names, then 'new and existing institutional and strategic investors.'

You're looking at a club placement; nobody's standing there as the conviction lead. They keep saying 'strategic' because they won't tell us the strategy. Here's what gets me - it's a public company. An 8-K is coming, so there is accountability on paper.

But the operating milestone? When does AI Factory number one light up? What's the take-or-pay? Silence.

Now a third geography is pouring capital into inference at scale, and nobody's defending the price floor. Here's Cristian Dina at The Next Web: Mirendil, founded by two researchers who left Anthropic after barely a year, has raised $200m at a $1bn valuation. The pitch: sell the self-improving AI that the big labs build for themselves and guard from everyone else. Mirendil - $200 million to build AI that improves AI.

Cristian Dina's got the byline at The Next Web, June 29. AI that improves AI. That's the whole sentence. $200 million for a loop.

And what drives me nuts: two hundred million dollars is a Series B-sized check riding on a one-line concept. Who's the paying customer? A model? Another lab?

Tell me where the revenue clears. Right after a $1.6 billion neocloud close, $200 million almost reads small. But I want the structure: is there a named lead setting terms, or is this another club where everybody writes and nobody owns the round?

If self-improving AI actually works, fine, it's the most valuable thing on the tape. If it's a research demo with a deck - $200 million is an expensive way to find out. Here's Straiker at Signalbase: The burgeoning field of AI-native security just saw a significant capital injection with Straiker announcing a $64.0M Series A funding round.

The startup which specializes in protecting agentic applications across every layer of the AI stack secured investment from a broad group of backers. So after a $1.6 billion mystery close and $200 million for AI that improves AI, here's the one on today's tape I actually buy. Straiker: $64 million Series A to secure the agentic workforce.

There's a real threat surface there. And this answers whether agentic security can get funded above seed. It's moving up the stack - Marathon and Illuminate are leading, and total funding is now $85 million. $85 million total means a $21 million seed before this.

That's lean for a company now pitching protection across every layer of the AI stack. This check is buying the build; they're not coasting on it. The strategic mix matters here - Workday Ventures and Citi Ventures are in alongside Bain Capital Ventures. That's three distinct enterprise buyers writing checks, which reads like procurement intent, not just sponsor money.

From Morningstar: Nebex the market infrastructure platform for the global space economy announced a $30M seed investment led by GV (Google Ventures) with participation from other top-tier venture funds. The company also announced a banking relationship with J.P. Morgan.

The capital raise and new banking relationship will help Nebex scale its platform and connect sovereign space programs with the founders and companies building new technologies to serve the space industry. Nebex - $30 million seed, and GV is the named lead. After the Sharon AI mystery we just hit, it's almost refreshing to have one check-writer with their name on the line. And a J.

P. Morgan banking relationship on a seed round. That's the part that catches me - institutional plumbing for revenue and cash flow before there's much revenue to plumb. The founder ran Axiom Space before this, so the space-economy angle has operator credibility.

Add a real lead and a clean cap table, and structurally, this is the cleanest round on the tape today. Clean cap table, sure. But 'market infrastructure for the global space economy' connecting sovereign programs to founders - the money's either buying engineering or buying legitimacy before the sovereign contracts are actually signed. I want to know which.

GV writing the lead check on a space fintech play with one conviction backer - compare that to the four-fund crypto syndicate we saw earlier this week with nobody willing to set terms. Same stage, completely opposite cap-table clarity. This one's from Fund Momentum: London-based Osney Capital has held the final close of its debut fund at a £60 million hard cap oversubscribed against an original £50 million target making it the UK's first venture firm dedicated exclusively to early-stage cybersecurity.

The British Business Bank anchored the vehicle with roughly £36 million with Imperial College London's endowment and Planet First Partners joining at final close. The fund writes pre-seed and seed cheques of £250k to £2.5 million and has already deployed into seven companies. Osney Capital closed its debut fund at a £60 million hard cap - oversubscribed against a £50 million target - and they're calling it the UK's first pure-play cyber seed fund.

The oversubscription line is true, but read the LP list before you call it pure market demand. The British Business Bank anchored at roughly £36 million - that's about 60% of the fund coming from one government-backed cheque. Sixty percent from the state? For a first-time manager, that's the whole ballgame.

After the Sharon AI piece we just hit, it's almost refreshing - at least here, I know exactly who wrote the cheque that mattered and why. And this is the unglamorous one I actually like - £250k to £2.5 million cheques seven companies already deployed and they only underwrite security. That's a regional specialist closing oversubscribed while everyone says only the mega-brands can raise.

GCHQ and NCSC talent have been spinning out with nowhere homegrown to pitch; somebody finally built the door. The catch is reserves. When 60% of your capital is one anchor LP, your follow-on math and recycling get complicated fast. The hard part of a debut raise isn't the first close - it's reserving enough to defend your winners through Series A.

If you track where startup dollars are flowing check out The Data Center Daily a daily briefing on AI compute hyperscaler capex the power grid semiconductor supply and energy markets. Find it wherever you listen to podcasts. What we’re watching next: Sharon AI’s forthcoming Form 8-K. It should add transaction details on the private placement and the convertible notes.

You’ll find links to every story we covered today in the show notes, so if one deserves a closer read, that’s the place to start. That’s Startup Fundraising for today. This is a Lantern Podcast.