PointFive. Say it out loud and you get it immediately — the company's entire value proposition is baked into a decimal point. Slash your cloud bill by half. Fifty percent. 0.5. That's the number they're selling, and after reading through what Accel just put behind them — $60 million in growth funding, led by the firm and joined by Index Ventures, Entre Capital, Perpetual Growth, Vesey Ventures, Sheva Ventures, and Salesforce Ventures — it's clear the market agrees.
The round brings PointFive's total raised to $96 million and its valuation to $500 million. Halfway to unicorn. The name isn't a coincidence. It's a promise.
New York-based PointFive built what it calls an "AI and cloud efficiency platform" that scans your entire cloud environment — AWS, Azure, GCP — through agentless, read-only integrations, then tells you where you're setting money on fire. No agents to install. No write access required. Just a complete picture of your AI infrastructure and a coach sitting on your shoulder saying, "Hey, you don't need that GPU running 24/7 for a batch job that happens at 3 AM."
Tokenmaxxing Is Real and It's Expensive
Here's the thing about enterprise AI adoption that nobody wants to admit at the all-hands: most of your employees are just playing with ChatGPT like it's a fancy search engine, and the bill is astronomical.
Accel partner Philippe Botteri has a name for it — "tokenmaxxing." You've probably done it. You fire off twenty queries to an LLM when one would do. Your team does it. Your whole company does it. Tokens are the units that track consumption of AI resources, and when everyone at an organization is maxing them out for marginal productivity gains, the costs compound faster than anyone in finance anticipated.
This isn't theoretical. Meta's CTO Andrew Bosworth sent a memo to employees back in April that went viral for its bluntness: "Nobody should be using AI tools just for the sake of using them. All motion is not progress and token usage alone is not a measure of impact of any kind."
Translation: stop wasting money on AI for the sake of looking innovative.
The problem is that most enterprises don't even know how much they're spending. The billing is fragmented across cloud providers, buried in dashboards that require a PhD to navigate, and tracked by nobody who cares about the bottom line. PointFive's whole thesis is that this has to change — and fast.
How the Platform Actually Works
PointFive doesn't rewrite your infrastructure. It doesn't touch a single configuration file. What it does is scan everything, understand what's running, and then make recommendations that cloud engineers can act on.
Chief Executive Aron Arvatz put it best when he told the Wall Street Journal: "We're their efficiency coach." The platform identifies wasteful spending on idle servers, unused storage resources, over-provisioned memory for AI models, and — perhaps most insidiously — "always-on" AI agents that drain finances around the clock without delivering proportional value.
One of the recommendations might be simple: use a cheaper model for this task. Not every customer query needs GPT-4. Sometimes a smaller, faster, cheaper model does the job just fine. PointFive figures out which tasks can be downgraded and which need the heavy lifting, then routes accordingly.
The company's "DeepWaste detection" identifies provider-specific optimization opportunities — from Azure OpenAI PTU rightsizing to AWS Bedrock inference optimization. It's not a one-size-fits-all approach. It learns your environment.
The Founders Know Waste Because They've Sold Through It
Arvatz founded PointFive in 2023 alongside Gal Ben David and Amir Hozez, cybersecurity veterans who previously co-founded IntSights Cyber Intelligence — a company acquired by Rapid7 for roughly $335 million in 2021.
Here's where the origin story gets interesting. During the integration of IntSights' technology into Rapid7's platform, the founders saw firsthand the scale of wasted cloud spending inside a mature organization. They watched millions burn on resources nobody was tracking, servers running for projects that had been abandoned, storage accumulating for data that served no purpose.
They left Rapid7 with a simple question: if this is how bad it gets at a company that size, what's happening at the startups and scale-ups racing to deploy AI?
The answer, apparently, is worse. Much worse.
The Numbers That Justify the Round
PointFive didn't disclose revenue figures, but Arvatz said annual recurring revenue increased six times over the past year. More telling: existing customers doubled their spending on average. They're paying more to save more — a counterintuitive dynamic that actually signals strong product-market fit. When your tool pays for itself and then keeps paying, customers naturally expand their engagement.
The customer roster reads like a who's-who of enterprise complexity: E.ON, the German utility giant; Nubank, Brazil's massive neobank; Fanatics, the sports merchandise and gambling platform. These aren't companies that can afford to leave money on the table.
Nubank's case is particularly striking. According to Arvatz, the neobank recouped its PointFive investment within ten days through cloud infrastructure cost reductions alone. Ten days. If that holds up as a pattern, the unit economics are almost absurdly favorable for customers.
Why Accel Cared Enough to Lead
Accel has a history of backing infrastructure plays that become table stakes — from early cloud computing bets to modern data infrastructure. PointFive fits that pattern precisely.
The timing is deliberate. Enterprise AI spending has gone exponential while governance frameworks have stayed linear, maybe even regressed. Companies pushed employees to adopt AI tools to "get more work done" without modeling what that would do to their cloud bills. Now the invoices are arriving, and they're shocking.
Botteri told the Journal that the cost of running AI applications has become one of the largest expenses for many organizations. Compute, networking, storage — all spiraling as companies scale AI adoption without corresponding cost controls. PointFive arrives at the exact moment this problem stops being manageable through spreadsheets and manual review.
The $60 million gives PointFive runway to expand into Europe and Israel, grow its engineering and marketing teams by roughly 40 people this year (down from an original plan of 80, because the company is using its own AI to avoid over-hiring — a bit of ironic self-application there), and launch new products like TokenShift, which debuted alongside the funding announcement to help customers track and control internal AI tool usage.
The Bigger Picture Nobody's Talking About
There's a pattern here that goes beyond any single funding round. The AI infrastructure market is splitting into two camps: those building models and those building the plumbing that makes models affordable at scale. PointFive is firmly in the second camp.
The cloud providers — AWS, Azure, GCP — offer basic cost tracking. But they're incentivized to sell you more compute, not less. They're the casino offering free drinks while you gamble. PointFive is the sober friend at the table telling you to check your stack.
This tension will only deepen as AI adoption accelerates. Every enterprise that moves from experimentation to production-scale AI will face the same reckoning: either you build cost discipline into your AI strategy from day one, or you'll end up like Meta — issuing internal memos telling employees to stop using AI tools just because they're fun.
PointFive's $500 million valuation suggests the market believes this problem isn't going away. Given what we know about enterprise AI spending trajectories, they're probably right.
