States expected to spend at least $45.6 million on initial technology changes tied to new Medicaid and food-assistance rules, according to company estimates and state documents reviewed by KFF Health News. Deloitte, Accenture and Optum were among the contractors positioned to perform the work.
The total is significant, but it needs a label. It combines estimates and early costs from five states with different systems and project scopes. It is not a verified measure of vendor profit, a complete national implementation bill or proof that each dollar had been awarded when the March 31, 2026 report appeared.
Five States Produced the $45.6 Million Floor
In Wisconsin, Deloitte estimated that two Medicaid work-requirement changes would cost nearly $6 million. Two related SNAP changes added $4.2 million. In Iowa, an Accenture estimate put Medicaid system work at roughly $20.3 million after an earlier estimate of up to $7 million.
Optum estimated about $1.8 million to assess and incorporate coverage restrictions in Vermont. Kentucky had already approved $1.6 million in initial changes to its Deloitte-operated eligibility system. In Illinois, Deloitte estimated at least $12 million for modifications, though the estimate also included work beyond the federal law. An Illinois official said a $2 million fixed fee was specifically related to work requirements.
Those figures sum to the reported floor, but they are not interchangeable. Some describe projected work, one covers an evaluation as well as implementation, and others refer to changes already approved. A defensible national total would require equivalent scopes, final payments and disclosures from every state.
The Rules Require New Eligibility Checks
Public Law 119-21 requires adults covered by the new Medicaid provision to document 80 hours a month of work or another qualifying activity unless exempt. KFF Health News reported that the requirement applies in 42 states and the District of Columbia. The law also moves millions of adults to eligibility checks twice a year and restricts coverage for several immigrant groups.
Eligibility platforms therefore need new data fields, business rules, exemption logic, notices and interfaces. That is the concrete reason for the technology work. The reporting did not establish that consultants were embedded in state agencies, transferring millions of individual records or controlling decisions outside their contracted systems.
The systems carry real administrative risk because eligibility errors can interrupt benefits for qualified people. But a history of failures in some systems does not prove that every new module will fail, that contractors intend exclusion or that any particular coverage loss will be caused by a software defect.
Federal Matching Funds Cover Much of the IT Cost
CMS guidance allows a 90% federal match for eligible design, development and installation costs in Medicaid eligibility systems. It provides a 75% match for qualifying maintenance and operations. States still need CMS approval and must show that commercial software choices are efficient and economical.
The same guidance calls for alternatives analysis, long-term ownership costs and consideration of dependencies that could lock a state into future procurement. Those conditions matter because federal participation is not a blank cheque or a finding that a quoted price represents value.
CMS said technology companies had pledged discounts or reduced rates through 2028. KFF Health News found that the details remained uncertain: state officials and companies did not identify consistent discounts across the reviewed projects. Deloitte said it was delivering its promised value; Accenture did not respond to repeated requests; Optum said it offered options intended to manage costs.
Coverage Projections Are Separate From Vendor Results
KFF Health News cited a Congressional Budget Office projection that the law's Medicaid policies would leave 7.5 million more people uninsured in 2034 than under the prior baseline. It also cited a 5.3 million increase in uninsured people specifically attributable to work requirements.
For SNAP, CBO estimated that changes in Public Law 119-21 would reduce participation by roughly 2.4 million people in an average month over 2025-2034. That is a reduction in participation, not a finding that 2.4 million people lose a fixed cash payment every month.
These estimates quantify expected policy effects relative to a baseline. They do not show how many people were removed by Deloitte, Accenture or Optum, and they do not attribute future disenrollment to software rather than the law's eligibility rules. Combining the projections with contract figures is useful context; converting them into a contractor-specific casualty count is not.
Revenue Is Documented, Profit and Performance Are Not
The investigation establishes that vendors will be paid to modify systems in response to a law expected to reduce enrollment. It also shows that federal taxpayers finance much of qualifying Medicaid technology work. That creates a legitimate oversight question about price, system accuracy and public accountability.
The evidence does not establish a coordinated preference for contractors over public employees, guaranteed revenue regardless of performance, or an intentional plan to deny eligible people. It also does not calculate margins, so contract value cannot be renamed profit.
The hard conclusion lies in the procurement record. Governments are imposing new reporting duties on systems already responsible for access to care and food aid, while key price and discount details remain incomplete. Every state should publish the final scope, federal and state shares, acceptance tests, error rates and payments. Without those records, calling vendors either indispensable experts or architects of exclusion substitutes ideology for performance evidence.