No tools for you
Hello, and welcome to Protocol Enterprise! Today: a summary of our recent reporting on the increasingly aggressive U.S. strategy to target chipmaking tools destined for China, how Intuit deals with automated tax return-reading software and a troubling pattern at SAP.
Kick ’em in the tools
Under the leadership of President Joe Biden, U.S. foreign policy related to China and tech has undergone a significant change. No longer is the administration content to attempt to keep Chinese tech a generation or two behind the U.S; instead, the White House is now attempting to damage its existing efforts.
To do so, the White House has elected to target semiconductor manufacturing equipment, under the theory that it can hobble China’s military efforts if it obstructs access to the expensive, complex machines needed to make chips.
- The administration’s approach revolves around the transistor, a fundamental building block of any logic chip that powers data center processors, PCs and smartphones.
- Specifically, the Biden administration plans to curtail access to chipmaking tools that can produce what’s known as a FinFET, or a fin field-effect transistor — referring to its physical shape when constructed at atomic scale.
- The Commerce Department issued an export control rule on software needed to design chips with a next-generation transistor called gate all around.
- Biden’s plan also blocks the chip equipment makers from servicing the tools they have already sold to the Chinese.
- The service and support for the systems is vital to ensuring they produce good chips, as efficiently as possible.
Whether the efforts to curtail China’s chip industry will succeed is another story entirely.
- Several former Commerce Department officials who worked on export controls told Protocol that in the long run, the U.S. must simply outpace China — the decision to hobble its chip production could help but isn’t a solution by itself.
- “What [China] is talking about is really kind of a direct challenge to American technology supremacy,” former Undersecretary of Commerce William Reinsch told Protocol recently. “And I think Biden's figured out the best way to meet that kind of challenge is to run faster. So that's the more important part of the equation, but tripping them along the way is OK in international politics and economics.”
- The $280 billion in chips and science funding that Biden recently signed into law are part of the U.S. plan to “run faster,” to use Reinsch’s foot race metaphor.
- Of that money, nearly $40 billion will flow directly to chip manufacturers to expand their U.S. capacity, but the legislative package includes billions in research and development funding, too.
China is not pleased with the U.S. plans. Part of the Chips Act prohibits any company that receives government money from building more advanced chip factories in China, and officials there have complained publicly about it.
- At a recent chip industry conference, Yu Xiekang, vice chairman of the China Semiconductor Industry Association, said that parts of the U.S. Chips Act discriminate against China.
- “We resolutely oppose the US’ restrictive actions targeting certain countries,” Yu reportedly said. “It contains essentially discriminatory clauses in market competition and creates an unfair playing field, which goes against the WTO’s fair-trade principles.”
- But China’s homegrown chip technology is insufficient to replace what the U.S. and other countries export.
- China’s government-led efforts have also bumped into problems. Several top executives associated with a state-run chip fund were recently arrested on corruption charges, throwing plans into chaos.
Sponsored content from DataRobot
DataRobot's AI Cloud for Financial Services Unlocks the Art of the Possible: DataRobot continues to attract clients in financial services who want to de-risk their AI investments and rapidly scale AI to almost every part of their operations, resulting in improved productivity and higher customer satisfaction.
Beating box 12 (and other feats of document understanding)
Box 12 was a tricky one.
Unlike other fields in the W2 tax form, box 12 can include one of nearly 30 different codes or be left blank altogether. That was tripping up Intuit’s optical character recognition system, which sometimes had trouble deciphering what tax filers using the company’s TurboTax software put in that space.
“We've actually built a model to just predict if something is blank, so that we know it wasn't a mistake — it actually really is empty. That's helped us increase our accuracy on that particular part of the document,” said Diane Chang, director of data science at Intuit.
Document understanding is getting better, Chang told me earlier this week. But there’s a lot more work to be done.
Right now, Chang’s team of data scientists are working to improve how Intuit’s Optical Character Recognition — which uses computer vision and natural language processing to interpret images and identify words in documents — understands the context of a document.
While these systems have gotten good at turning image data into text, they are now just beginning to make sense of text in relation to a document’s purpose or broader context, Chang said.
“This structure of documents has so much information that you don't want to lose that structure and just sort of turn it into a bunch of text,” Chang said.
Around the enterprise
Two female SAP employees said the company’s HR department failed to respond effectively to reports that they were raped at company events, according to a report from Bloomberg.
The LockBit ransomware group said it was behind an attackearlier this year on Entrust, a security vendor.Sponsored content from DataRobot
DataRobot's AI Cloud for Financial Services Unlocks the Art of the Possible: Banks need to secure a competitive advantage in an increasingly tight race to harness best-in-breed technology. Decision makers need to not just plan a future-ready strategy, but also recognize the value of AI that could boost not just their performance in-house but also their reputation among their global customers.
Thanks for reading — see you Monday!
Recent Issues
In a tough economy, benefits of the cloud 'only magnify'
November 14, 2022
Twitter’s security leads just quit. Now what?
November 10, 2022
Intel finally serves up a chip
November 09, 2022
The great AI race that wasn’t
November 08, 2022
Cloudflare sets a target
November 07, 2022
How Elon will bring back the Fail Whale
November 04, 2022
See more
To give you the best possible experience, this site uses cookies. If you continue browsing. you accept our use of cookies. You can review our privacy policy to find out more about the cookies we use.