NVIDIA just created a new category for neoclouds. Now what?

For years NVIDIA's data centre line told one story: the hyperscalers were buying everything. This earnings report it split in two: Hyperscaler, and a new category called ACIE - AI Clouds, Industrial and Enterprise. The second line is the one that matters if you're running marketing for a neocloud.

ACIE is a bucket. It groups neoclouds with the enterprises buying from them and the industrial customers running AI on their own kit. NVIDIA didn't draw that line for accounting reasons - they drew it because the demand inside the bucket is now large enough to need its own number. The story behind that number is straightforward: a lot of enterprises want their own AI infrastructure - sovereign, specialised, in their region, on their data - and most of them can't or won't build it on the hyperscalers. They need specialist providers. That's you.

This is a starting gun. Enterprise buyers are at the beginning of a multi-year build-out. They are starting to look. The RFPs haven't all landed yet, but the shortlists are forming, and most of them are forming in places that don't look anything like a traditional buying cycle. The neoclouds that show up now - specifically, technically, consistently - will be on those shortlists. The ones that wait will be reading about them.

The unhelpful thing is that everyone else got the same memo. Every neocloud, every regional infra player, every "sovereign AI" provider with a comms team is about to claim ACIE leadership. Most of them will sound the same: vague enterprise-grade, a stock photo of a server rack, a quote about partnership. The differentiation problem just got harder, fast.

Visibility in the era of zero-click search

The way buyers find providers has changed and isn't changing back. Enterprise CTOs, heads of ML platform and procurement leads aren't sitting through a ten-page Google search journey to your homepage anymore. The decision happens earlier and elsewhere - in the snippet they read without clicking, in the AI answer that synthesises the market for them, in the LinkedIn post a peer reshares, in the podcast they had on during the commute, in the analyst note that landed in their inbox.

That's the awkward shift for marketing leaders who grew up measuring traffic. The shortlist often forms before there's any traffic to measure. Which means the answer to "how do we get found?" is no longer "rank for the right keywords." It's "show up everywhere those buyers form opinions" - and the channels that actually shape opinion in enterprise AI right now break down roughly into four.

• Earned media. Specialist trade press, tier-one tech outlets when the story warrants, regional business press for sovereign-AI plays. Earned coverage is the third-party authority signal that AI answer engines, analysts and procurement teams all weight heavily. A single piece in the right specialist outlet outperforms a dozen in the wrong ones.
• Influencer relations. Not consumer influencers - the practitioners enterprise buyers actually listen to. The podcast hosts, the Substack writers, the GitHub-active engineers, the analysts and ex-CTOs who shape developer and infrastructure opinion. A mention or guest spot in the right one moves perception faster than any campaign.
• Social media. LinkedIn is where enterprise architecture decisions get gossiped about before they happen. Your engineers posting publicly about real workloads, your CEO showing up in the right conversations, your customers (when they can) talking about why they chose you. Owned distribution that doubles as social proof.
• AI visibility. The synthesis layer. ChatGPT, Gemini, Claude and Perplexity don't invent their answers - they pull from everything above. Your presence here is largely a downstream signal of how well the other three are working, plus the technical and structural decisions that make your content easy to cite.

The point isn't to pick one. It's that they compound. Trade coverage gets quoted by influencers. Influencer commentary gets resurfaced in social. Social conversation gets cited in AI answers. Each channel makes the others work harder. A comms programme that only invests in one is leaving most of the visibility on the table.

How to stand apart from the other neoclouds

You can't out-NVIDIA NVIDIA. You can't out-AWS AWS. But you can be unmistakable on the dimensions enterprise AI buyers actually care about. A few things worth investing in across all four channels:

• A real technical voice. Engineering blogs that show your work. Benchmarks you'll let other people reproduce. Failure post-mortems written like you respect the reader. This is the raw material for everything else: it's what specialist press will cover, what podcast hosts will invite you on to discuss, what your engineers can post about on LinkedIn, and what AI answer engines will cite. Marketing copy that floats above the technical detail gets skipped in every channel.
• A lane you actually own. The neoclouds that will matter in 2027 aren't the ones claiming they do everything. They're the ones who own a specific lane - LLM inference for life sciences in the EU, real-time AI for ad-tech in APAC, sovereign workloads for the Gulf, fine-tuning for regulated industries. Pick the lane and become unignorable in it across press, social and the practitioner conversations in it.
• Third parties saying it for you. Analyst notes, specialist outlets, podcasters in your category, Substack writers who shape developer opinion, conference keynotes that aren't sponsored. A single MLPerf result or a mention from a respected practitioner does more for a neocloud's credibility than ten press releases, in every channel that matters.
• Visible reliability. Enterprise infrastructure is a trust purchase. Status pages, SLA history, post-incident transparency, named customer references. These signals show up in trade coverage, in social conversation, and inside AI answers when buyers ask comparative questions.
• Supply chain candour. H200s, B200s, GB300s, transformer constraints, power siting. Enterprise procurement teams are quietly obsessed with this stuff. The neoclouds that talk openly about how they secure capacity, where their data centres draw power from, and what their roadmap is for next-gen silicon get pulled into RFPs the others don't - and the same candour earns coverage in the trades and credibility on social.

None of that lives inside a press strategy alone. It's content, engineering, customer success and comms working as one motion - because that's what enterprise buyers are evaluating when they choose.

What's worth measuring

Clip counts and AVE will tell you nothing useful about ACIE. A handful of cross-channel signals will:

• Specialist citation density (earned). What share of the trade and analyst coverage in your category names you, versus your closest two or three competitors? Density matters more than volume - five citations from authorities your buyers trust beat fifty from outlets they don't.
• Practitioner mentions (influencer). Are the right podcasters, Substack writers and engineers talking about you unprompted? Quarterly tracking of named mentions in the practitioner channels your buyers consume.
• Social share of voice (social). How often is your brand and your competitors' brand surfacing on LinkedIn in your category, and what's the quality of the conversation - your engineers leading it, or your competitors'?
• AI answer presence (AI visibility). If a CTO types "best neocloud for LLM inference in [your geo]" into ChatGPT, are you named, ranked and described correctly? If not, that's the gap.
• Commercial pull-through. Are inbound RFPs, trial sign-ups and warm intros tracking with the narrative across the four channels? Rising visibility without rising RFPs means you're talking to the wrong audience or describing yourself wrong. Both rising together means you've found something to scale.

The window won't stay open

Most neoclouds will treat ACIE as a slide in next quarter's all-hands and move on. A handful will treat it as a marketing event - the moment to anchor a year of cross-channel positioning around a category that's about to be in every enterprise architecture review.

The bet is straightforward. Enterprise AI is moving from experiment to production over the next 12-36 months. The work will be split between hyperscalers and the specialist providers buyers can find, trust and reason about. The neoclouds that compound their visibility now - in earned coverage, in practitioner conversations, on social, and inside the AI answers that synthesise all of it - own the shortlist. The ones that wait for the RFP to arrive before telling their story will be on the long list, if they make a list at all.