New Jersey Cut Phone Waits from 40 Minutes to 90 Seconds

Nathan Labenz opens Cognitive Revolution slightly bewildered. He spends his days inside the AI bubble and is often frustrated by how few business and government leaders will grapple with how large the next few years could be. Then Governor Phil Murphy's team reached out and asked to do the episode — a first, in Labenz's experience. Using OpenAI's deep research on Murphy's record, Labenz concluded that the time, energy, and desire to learn Murphy is bringing to AI put him among a small set of American elected officials, especially for someone in the last year of a term-limited tenure who could easily have coasted.

The concrete scoreboard already looks unlike most statehouses. New Jersey has trained more than 13,000 state employees on AI tools, used systems to cut waits for government services — in one case from more than 40 minutes to 90 seconds — and established an AI Hub with Princeton University, with Microsoft and CoreWeave as corporate partners. The hub is meant to promote R&D, train workers on new tools, and push deployments that benefit all residents. Labenz also flags the harder file still open: efficiency versus equity as jobs shift, energy versus a 2035 clean-power target, whether states should regulate if Washington will not, and whether U.S.–China cooperation is still imaginable.

When Labenz asks how Murphy came to say "everything is on the table," and how many other electeds are with him, Murphy is careful on the last point. He is all in, but AI does not come up a lot in discourse with other elected officials. Inside New Jersey he deals with the congressional delegation, counties, locals, and the legislature; outside the state he mostly traffics with other governors. In that circle, he says, it is not a hot topic.

The click, Murphy says, came about three years ago at a seminal dinner in the Bay Area. He was there with his wife, a full-time volunteer for the state, and several colleagues, sitting with significant AI thought leaders, most of them running businesses. Mustafa Suleyman was one of them; Murphy has kept in touch. The table spent a while commiserating about cost of living, regulation, and the Bay Area business environment. Murphy asked the obvious follow-up: after that litany, why are you all still here? The answer was talent.

On the way out he ran New Jersey's numbers. It has, in his telling, the number-one public education system in America, Princeton and Rutgers, and the highest concentration of scientists and research professionals per square mile in the world, plus an innovation legacy that includes Einstein, Edison, and Bell Labs. There is no reason, he concluded, the state cannot stake a claim in this emerging realm. He insists on the word *realm*, not *technology*. New Jersey is a big pharmaceutical and life-sciences state; the whole notion of drug discovery is going to be transformed. The mistake a lot of people make, he says, is parking AI in a narrow technology lane. It is going to change all of our lives.

Labenz asks what actually made it click, because people in AI are often bewildered that people in power will not treat this wave as bigger than the last ones. There was no demo that night, Murphy says, though they have had demos since then that have blown them away. What poisoned the well, in his view, was crypto. He is not a "hell no" crypto person, but the so-called use case has never jumped out at him, and a room of early-stage companies a buddy gathered in New York left him with enormous frustration. Generative AI is the opposite: use cases overwhelming, impact obvious. If you take a snapshot of public discourse you might be disappointed relative to the potential; if you film a video, he thinks the curves start to converge as more people see it with their own eyes.

Labenz pushes on the snapshot-versus-video comfort. Sam Altman and Anthropic's Dario Amodei have been saying, pretty clearly, that superhuman AI able to do everything better than humans may arrive around 2027. Murphy may be right that the lines are converging, Labenz says, but maybe not at the pace they should. Does that radical possibility weigh on a sitting governor, given that the people building the systems are saying it is perhaps two years out? Murphy will not argue with the geniuses at the forefront. He spends his time on pragmatic implications, and he grants the fair point: the lines may not be converging as fast as they should.

That is the frame for the New Jersey AI Hub, announced with Princeton a little over a year earlier. A week or two before taping, Microsoft and CoreWeave — a hyperscaler born and bred in New Jersey — came in as two key founding corporate pillars. Murphy's line is the one Labenz teed up in the cold open: he is not going to be the guy who takes us to superhuman intelligence, but he is assuming it will happen. He wants New Jersey to have the very best catcher's mitt when all of that comes their way.

The mitt is a place, not a slogan. He wants residents who need a government service, startups that want into the Princeton venture, and thought leaders wrestling with ethics and regulation all to lean the same way on the margin: a lot of that can be found inside the four walls of New Jersey. Training the state workforce is free, and the state offers its AI assistant package to other governments, other states and cities. "We're all in this together in that respect," he says.

After the break, Labenz asks who actually uses the New Jersey AI assistant. It is built on Microsoft Copilot. Murphy jokes that he is the least qualified guy to be using it, but about 13,000 of roughly 60,000 state employees use it regularly, and everyone is trained for free. Credit, he says, goes to chief AI strategist Beth Noveck, with him since day one after running the state's innovation office; Dave Cole, who succeeded her on innovation; CTO Chris Ryan; and the departments where the users actually sit. Noveck later gets a fuller introduction: she advised David Cameron, Barack Obama, and Angela Merkel before landing on Murphy, a trajectory he still finds slightly mysterious.

Surveys in both the public and private sector show the split he lives with: enthusiasm to learn, and rightful concern about whether the job will still exist in X years. His own observation, in year eight as governor, is that everyone including him is stuck too often in mundane weeds. Important work is 20 percent of life when it should be 80. AI, early days and not to be overstated, is starting to reverse that. Upskilling is no longer a fancy consultant word; it is actually happening. Residents are happier in the obvious places — call centers, unemployment insurance, motor vehicles — because answers come faster.

Labenz asks for numbers. A colleague's notes: a 50 percent boost in call resolution rates through self-service options, and for one program, phone wait times from 40-plus minutes to an average of 90 seconds. Murphy's monthly television and radio show, *Ask the Governor*, is his reminder that this is what 9.5 million residents actually call about. They do not mostly want a take on immigration; they have been on the phone for hours trying to get an unemployment benefit resolved, or nobody is answering a property-tax question. It is Tip O'Neill's "all politics is local" applied to government: one foot in front of the other, make it work. Labenz's verdict is that the state is ahead of many companies that are still in meetings about whether to try.

Labenz floats two extensions of what citizens might rightfully expect: if you are entitled to a lawyer, should you also be entitled to AI to think through your case; and should local jurisdictions prepare for self-driving instead of waiting to be hit by it. Murphy had already been listing new surfaces. The Office of the Public Defender is building an AI-powered brief bank so defenders can draft legal briefs more efficiently — they are not in it for the money, so anything that speeds the process matters. Business certifications, permitting, and notary work are on the list too. On the criminal-justice question he had not thought about an AI entitlement; he calls it a fair one.

On autonomy his view is settled: self-driving is here to stay, like a lot of things, but it has to be done in a way that is equitable to people who make their living from driving. The New York–New Jersey port is one of the largest in the world, and probably 90 or 91 percent of the traffic is on the Jersey side, which means enormous trucking and ancillary livelihoods. New Jersey is the most densely populated state in America. Every time he goes to Arizona he sees Waymos everywhere, and he has told the Waymo people that if they really want to get this right they should test in Jersey City or Hoboken, where there is real traffic.

He would bet the biggest impacts, in Jersey and around the world, will be in the private sector simply because of scale. State GDP toggles between $700 and $800 billion; the state budget he is about to unveil is in the high $50 billions. A Paris AI gathering in the days before taping was enlightening and, at some level, concerning; how this plays out at the national and geopolitical level probably dwarfs everything else in the conversation. On the efficiency-versus-jobs trade inside government, he does not think they are anywhere close. Unmet demand is still larger than generative AI's reach. If headcount ever falls, he expects attrition — not hiring the next body — rather than a one-day cut. The follow-on is the hard part: where do the people you never hired go? "That will be very, very tricky." Employment and operating businesses in the state are at historic highs; he does not see the trade as tomorrow, 2027, or 2028.

Labenz turns to data centers and the energy to power them, including tension with local opposition and New Jersey's 2035 target of 100 percent clean energy. Murphy splits the jobs into a barbell. On one end are CoreWeave-style hyperscalers: big compute, big energy, headquarters thankfully in New Jersey, but data centers are not a big employment spinner. On the other end are talent-dense AI companies, the kind he visits in California, which remind him of the old Goldman Sachs bullpen — open architecture, everyone all-in, certified geniuses. Princeton, he thinks, will be a huge magnet for that second group, at least on the East Coast. New Jersey sells well internationally and treats California as a nation-state: Los Angeles for film and television, the north for technology and, these days, AI.

Energy is the harder ledger. He regrets that they have not found a compromise with the Trump team on offshore wind, where New Jersey had the largest program in America and has had to hit pause to look for common ground. Significant nuclear power remains certified in some cases through the middle of the century, including what he thinks is one of the largest remaining campuses in America in South Jersey. He views that as carbon-neutral, knowing not the entirety of the environmental community agrees. CoreWeave is building out in New Jersey and, so far, has not tripped a dirty-energy wire. "So far we've been able to have our cake and eat it too." Other firms are kicking the tires as well.

Land is scarce — ninth, tenth, or eleventh in population, first in density — so the deals are one-offs and have to be creative. A former pharmaceutical campus in Kenilworth, after the drugmaker moved elsewhere in-state, was refreshed by a developer and has struck an arrangement with a big AI player Murphy will not name because it may not yet be public. The state was right behind it, including a check that the energy situation was sufficient. Plopping a data center on or near a nuclear campus is another potential creative solution. If the industry keeps going this direction, and if DeepSeek is interesting but does not materially change the trajectory — if he is wrong, all bets are off — he thinks it will lead to a renaissance of nuclear technology and nuclear power, whether or not that renaissance is in New Jersey.

On China, Murphy is speaking from a former ambassador's file: he has lived in China, done business there, lived in Hong Kong, dealt with their missions overseas, and watched how they approach intellectual property even inside New Jersey, a massive per-capita research state. Privacy is not an issue in that system; the collective matters at the expense of the individual; that has always been the MO and always will be; AI will only further all of the above. As a nation he thinks the United States should be proud of its general technology leadership, a badge of honor not limited to generative AI. Then the hard line: nobody is going to hand us the title, and there is no more compelling entity in that race than the People's Republic. God willing, America finds its footing on its own values. He will not assume that happens. Taping-day aside, he notes that American values have not had a good three and a half weeks.

Labenz asks for himself what ordinary people can do to build bridges rather than ratchet up an arms race that ends badly for everyone. Murphy agrees the contest will not be Eagles 34–0 in the third quarter of the Super Bowl: there will be a first-place winner and a second-place winner, not a shutout. Advice for dealing with Chinese entities is scarce. He made friends and contacts there over the years and remains a big skeptic on intellectual-property theft. Institutionally, he fears there are few constructive ways forward for "guys like you and me." Labenz calls it sobering; Murphy says he is giving it as he sees it.

On state-level AI law he gives a to-be-determined. A month earlier he thought regulation should be federal; after the election he is less sure what that looks like, and he had shared worries that the Biden team had gone too far restricting access. Europe, in the earlier fairness discussion, may have gone too far too early without having the answers. California came close to doing something and pulled back, with big potential fallout on both sides. What New Jersey will clearly do is make sure that as AI becomes more pervasive, its use is consistent with laws already on the books — anti-discrimination and other basic pillars. A federal solution is still the strong preference. Closing, he enjoys the conversation enough to want Labenz connected with Beth Noveck, and he offers a prediction Labenz had flagged as rare situational awareness: in three months, maybe six, this conversation will look stale. They should do it again. The distance traveled from ChatGPT to today will be dwarfed by the distance from today to two years from now. He will be frustrated not to be governor — only 11 months left — and wants to stay in the space. The takeaway from talking to Labenz: you cannot let any grass grow under your feet.