If any vertical industry should have been immune from public cloud computing, it’s government. Risk-averse and not particularly innovation or cost-driven, governments are most likely to play it safe with legacy vendors and legacy approaches to enterprise IT.

But that’s not what’s happening. At least, not completely.

While plenty of roadblocks remain to government cloud adoption, various states are showing the US federal government how to innovate and cut costs with the cloud.

The federal cloud spirit is willing…

The tipping point for the federal shift to public cloud computing came from then US CIO Vivek Kundra, who declared a goal of shifting 25% of the federal government’s $80 billion IT budget to the cloud.

Though greeted with euphoria, the reality has been somewhat muted.

As the US Government Accounting Office (GAO) found in a 2014 report, federal agencies increased the total number of cloud computing services implemented by 80 services, from 21 to 101. At the same time, they spent $222 million more on cloud services, going from $307 million to $529 million.

That’s the good news.

The bad news, however, is pretty bad:

“Even though the agencies collectively and individually increased the percentage of their IT budgets allocated to cloud services, our analysis showed that the agencies are still devoting a large portion of their IT budgets to non-cloud computing expenditures. Specifically,the agencies in 2014 were collectively budgeting 2 percent of their IT budgets to cloud services, while the remaining 98 percent were dedicated to non-cloud expenditures.”

The reason for this slight improvement in cloud spending comes down to inertia, for the most part: “these agencies collectively had not considered cloud computing services for about 67% of their investments” because they “had only planned to consider cloud options for [cloud] investments when [systems] were to be modernized or replaced.”

This wasn’t what Kundra wanted, but it’s what happened. Agencies cite difficulty of meeting federal security requirements (anticipated and met by AWS and others), expertise, and other factors as reasons for their failure to meet targets.

Still, there are some bright spots within the federal government.

Bright clouds in the federal government

First, despite the dire outlook at the federal level noted above, more innovative agencies are forging ahead with public cloud, including General Services Administration (GSA) and the Department of the Interior.

The biggest (known) cloud deal to date is the CIA’s $600 million contract with AWS. While legacy vendor IBM fought in court to block that deal, AWS ultimately prevailed. It has since also won over the US Department of Defense, taking on sensitive data (and reducing other agencies’ excuses about cloud security).

This may be precisely what the US government needs.

The federal government has already demonstrated its ability to fail at the cloud (think: Healthcare.gov). Given that the GAO highlighted cloud expertise as one of the biggest barriers to success, letting AWS build the cloud is a big deal.

By giving control to the cloud leader, Amazon, the government’s odds of success have gone way up.

States show the way to the cloud

But as impressive as some cloud wins are at the federal level, it’s really the state governments that shine.

For example, Kentucky CIO Kent Fowler insists, “I’m a strong proponent that government will be out of the infrastructure business. We can’t afford it. In the next five to seven years, I suspect that 80-plus percent of our infrastructure will be off our premises and in the cloud somewhere.”

80%+? That’s better than the private sector, and it’s also much better than the federal government.

Of course, not every state is at Kentucky’s level of commitment, but states as diverse as Alabama, Utah, and North Carolina are all jumping into the cloud.

Sometimes, this is a matter of buying into infrastructure-as-a-service, and sometimes it’s a more pedestrian move to software-as-a-service. Sometimes, as in Ohio, it’s a matter of consolidating infrastructure into a meta-cloud across state agencies.

The results can be impressive. California, for example, has managed to decrease its storage footprint by 30%, even as it has increased storage space by 300%, all while shaving 35% in costs across 29 agencies.

However it’s done, it’s getting done. So, yes, while the US federal government writes bigger checks, the states may be the real laboratories of government cloud innovation.