Nancy Pelosi signing the Chips Act
Photo: Kent Nishimura/Los Angeles Times via Getty Images

Chipping in from the rough

Protocol Enterprise

Hello and welcome to Protocol Enterprise! Today: After the chip industry won the lottery this week, the focus turns to the breadth and depth of its U.S. manufacturing plans, on-call AI ethics, and the cloud market continues to drive the enterprise tech train.

Be careful what you wish for

This week the chip industry finally got its $52 billion to subsidize manufacturing in the U.S. Getting there wasn’t easy, or cheap: The money was attached to a lobbying bill of over $100 million and it took two years of Congressional squabbling.

As recently as a few weeks ago, the effort looked doomed. But now that Congress has spoken, the industry must deliver.

  • The largest four chip manufacturers have all announced various plans to expand in the U.S. or build new facilities altogether, and other companies have openly discussed additional U.S. expansion plans
  • New chip factories are not so easy to build, and take at least two or three years to complete, and potentially more time to ramp up to the high volume of throughput necessary for modern chipmaking.
  • TSMC has said its Arizona facility will open in 2024; Intel says 2025 for its new Ohio fab; Samsung says it will ramp to high volume in the second half of 2024.

Building the fabs is one thing. But without the appropriate chipmaking tools, the factories are useless.

  • Up until recently, semiconductor manufacturing equipment was in short supply, like everything else — including, ironically, a shortage of chips required for the tools that make the chips.
  • As recently as Wednesday, chip equipment maker Lam Research said on its earnings call it continued to struggle with supply chain and component issues, though the situation was improving.
  • ASML, which exclusively makes EUV machines, is set to produce only 55 machines, with plans to ramp up to 90 in the longer run. That’s in sharp contrast to the 400 machines using earlier-generation technology.

The Commerce Department is going to be responsible for doling out a lot of the cash. Officials publicly say that the agency is up to the task and eager to get started.

  • Undersecretary of Commerce for Standards and Technology Laurie Locascio said in a recent interview with Protocol that Commerce has been working for more than a year to build up the necessary personnel and other infrastructure.
  • “How do we do this?” Locascio said. “How do we get the money out the door? How do we hire the people? All of that planning has been going on for the past year.”
  • Locascio says that beyond the pure logistics, staffers are also considering the best way to strategically allocate the funds to achieve the objectives set out by Congress.
  • But staffers Protocol spoke with are less confident in its inevitable success, and say that privately the concern is mismanaging public funds.

Real the full story here.

— Max A. Cherney (email | twitter)

Dial-an-AI-ethicist?

With more and more off-the-shelf machine learning models, open-source large language models and no- and low-code software available today, any small business can inject AI into products — sometimes even if they don’t have a data science dabbler on staff. But even when companies have more deliberate AI goals and an engineering team, they may not have people dedicated to assessing the potential impacts of algorithmic systems (other than what they mean for the bottom line).

Does that mean companies should subcontract out their AI ethics?

Northeastern University thinks so. The school’s Institute for Experiential AI just launched an independent AI Ethics Advisory Board. The group of 40 people from academia and industry will provide external consulting on questions related to AI ethics.

Consulting giants including Deloitte and KPMG already offer this type of service. And startups have emerged to help assess whether machine learning models are built in a way that could produce discriminatory decisions or drift from their original intent.

As tech giants like taser maker Axon and Google feel the burn of reputational backlash when internal AI ethics approaches go awry, more startups might opt for outside ethical help. But accountability, transparency, explainability and other AI ethics cornerstones should not be mere compliance boxes to check. Hopefully when companies hire AI ethics contractors it leads to baked-in results rather than the thin icing of an ethics-washed PR win.

— Kate Kaye (email | twitter)

Cloudy forecast

The overall cloud infrastructure services market — IaaS, PaaS and hosted private cloud — grew 29% worldwide to $54.7 billion in the second quarter despite deteriorating economic conditions and the impact of the strengthening U.S. dollar on foreign currency exchange rates, according to a new Synergy Research report released Thursday. Excluding the effect of those rates, the cloud market grew approximately 35%.

Industry leader AWS saw its cloud market share increase by a point to 34%. AWS revenue jumped 33% to $19.74 billion for the three months that ended June 30, and Amazon reported a $100.1 billion backlog of customer contract commitments for future services that are primarily related to AWS.

AWS and rivals Microsoft and Google combined for a 65% share of the worldwide cloud infrastructure market in the quarter, up from 61% in the same period last year.

Microsoft’s cloud market share dropped a point to 21%, according to Synergy. Microsoft Cloud revenue — which includes Azure and other cloud services, Office 365 Commercial, the commercial portion of LinkedIn, Dynamics 365 and other commercial cloud properties — exceeded $25 billion for the first time with 28% growth. Microsoft reported a $193 billion backlog of customer commitments, $189 billion of which is attributed to that division.

While Google Cloud saw a “meaningful uptick” in its cloud market share, it did not gain a full percentage point and remained at 10% in Synergy’s report. Alibaba Cloud (5%), IBM/Kyndryl (4%), Salesforce (3%), Tencent (3%) and Oracle (2%) also held steady from the first quarter.

Google Cloud revenue, which includes Google Cloud Platform, Google Workspace collaboration tools and other enterprise services, for the first time eclipsed $6 billion last quarter. Parent company Alphabet reported a $51.2 billion revenue backlog primarily related to Google Cloud and approximately half of which it expects to recognize in the next 24 months.

— Donna Goodison (email | twitter)

Around the enterprise

Summer Fridays, right? SAP has partnered with Coldplay’s Chris Martin to promote its sustainability efforts, which must have thrilled a couple of elder millennials in SAP’s marketing department.

Thanks for reading — see you Monday!

Recent Issues

Follow the data